Relating Pupil Diameter and Blinking To Cortical Activity and Hemodynamics Across Arousal States

bioRxiv preprint doi: https://doi.org/10.1101/2022.06.15.496309; this version posted October 31, 2022.
The copyright holder for this preprint

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1 Title: Relating pupil diameter and blinking to cortical activity and hemodynamics across arousal states
2 Abbreviated title: Neural and hemodynamic correlations with pupil dynamics
3 Authors: Kevin L. Turner1,2, Kyle W. Gheres2,3, Patrick J. Drew*1,2,3,4
4 Affiliations:
1
5 Department of Biomedical Engineering, The Pennsylvania State University, University Park, PA 16802
2
6 Center for Neural Engineering, The Pennsylvania State University, University Park, PA 16802
3
7 Department of Engineering Science and Mechanics, The Pennsylvania State University, University Park,
8 PA 16802
4
9 Departments of Biology and Neurosurgery, The Pennsylvania State University, University Park, PA
10 16802
11 *Corresponding author: Patrick J. Drew - pjd17@psu.edu
12 Acknowledgements: We thank Nikki Crowley for comments and feedback on the manuscript. This work
13 was supported by NIH grants R01NS078168 and R01NS079737 to P.J.D.
14 Conflict of Interest: The authors declare no competing financial interests.
15 Author Contributions: K.L.T. and P.J.D. designed the experiments. K.LT performed experiments.
16 K.L.T., K.W.G., and P.J.D. analyzed the data. K.L.T. and P.J.D. wrote the paper.
17
18 Abstract (250 words)
19 Arousal state affects neural activity and vascular dynamics in the cortex, with sleep associated with
20 large changes in the local field potential (LFP) and increases in cortical blood flow. We investigated the
21 relationship between pupil diameter and blink rate with neural activity and blood volume in the
22 somatosensory cortex in male and female unanesthetized, head-fixed mice. We monitored these variables
23 while the mice were awake, during periods of rapid eye movement (REM), and non-rapid eye movement
24 (NREM) sleep. Pupil diameter was smaller during sleep than in the awake state. Changes in pupil
25 diameter were coherent with both gamma-band power and blood volume in the somatosensory cortex, but
26 the strength and sign of this relationship varied with arousal state. We observed a strong negative
27 correlation between pupil diameter and both gamma-band power and blood volume during periods of
28 awake rest and NREM sleep, though the correlations between pupil diameter and these signals became
29 positive during periods of alertness, active whisking, and REM. Blinking was associated with increases in
30 arousal and decreases in blood volume when the mouse was asleep. Bilateral coherence in gamma-band
31 power and in blood volume dropped following awake blinking, indicating a 'reset' of neural and vascular
32 activity. Using only eye metrics (pupil diameter and eye motion), we could determine the mouse’s arousal
33 state (‘Awake’, ‘NREM’, ‘REM’) with greater than 90% accuracy with a 5 second resolution. There is a
1
bioRxiv preprint doi: https://doi.org/10.1101/2022.06.15.496309; this version posted October 31, 2022. The copyright holder for this preprint
34 strong relationship between pupil diameter and hemodynamics signals in mice, reflecting the pronounced
35 effects of arousal on cerebrovascular dynamics.
36 Significance Statement (117 words)
37 Determining arousal state is a critical component of any neuroscience experiment. Pupil diameter and
38 blinking are influenced by arousal state, as are hemodynamics signals in the cortex. We investigated the
39 relationship between cortical hemodynamics and pupil diameter and found that pupil diameter was
40 strongly related to the blood volume in the cortex. Mice were more likely to be awake after blinking than
41 before, and blinking 'resets' neural activity. Pupil diameter and eye motion can be used as a reliable, non-
42 invasive indicator of arousal state. As mice transition from wake to sleep and back again over a timescale
43 of seconds, monitoring pupil diameter and eye motion permits the non-invasive detection of sleep events
44 during behavioral or resting-state experiments.
45
46 Introduction (634 words)
47 The dynamics of the eyes convey information about mental state. Besides their respective roles in
48 controlling light levels and protecting the eye, pupil diameter and blinking give information about the
49 state of neural activity in the brain (Strauch et al., 2022). However, for measures of pupil diameter and
50 blinking to be useful, we must understand their relationship to neural and vascular physiology. Pupil
51 dilations are associated with higher levels of arousal during the awake state (Drew et al., 2001; Hess and
52 Polt, 1964; Kahneman and Beatty, 1966; Morad et al., 2000; Onorati et al., 2013; Yoss et al., 1970) and
53 correlate with increased sympathetic activity (Bradley et al., 2008). The pupil will usually dilate when a
54 subject is performing a task or making a decision (Burlingham et al., 2022; Einhauser et al., 2010;
55 Gilzenrat et al., 2010; Hakerem and Sutton, 1966; Nassar et al., 2012) and recent rodent work has shown
56 that pupil diameter fluctuations temporally track cortical state (McGinley et al., 2015; Reimer et al., 2014;
57 Vinck et al., 2015). In both rodents and primates, pupil dilation reflects noradrenergic tone in the cortex
58 as well as activity in the locus coeruleus (LC) (Joshi et al., 2016; Larsen and Waters, 2018; Preuschoff et
59 al., 2011; Reimer et al., 2016). In addition to its effects on neurons, noradrenergic input from the LC
60 causes vasoconstriction of cortical arteries (Bekar et al., 2012). LC activity and noradrenergic tone
61 provides a potential mechanism for coupling arousal to cortical hemodynamics (Pisauro et al., 2016), as
62 changes in pupil diameter are also correlated with blood-oxygen-level-dependent (BOLD) signals seen in
63 neuromodulatory centers during functional magnetic resonance imaging (fMRI) (Pais-Roldan et al., 2020;
64 Sobczak et al., 2021). Both head-fixed and freely behaving mice frequently sleep with their eyes open
65 (Karimi Abadchi et al., 2020; Senzai and Scanziani, 2022; Turner et al., 2020; Yuzgec et al., 2018). Pupil
66 diameter decreases during sleep and can be used to detect sleep events on a minute-to-minute timescale
67 (Karimi Abadchi et al., 2020; Yuzgec et al., 2018), making it a particularly useful metric for attention and
2
68 behavioral monitoring.
69 Blink rate is influenced by mental state and fatigue (Holland and Tarlow, 1972; Nakamori et al.,
70 1997; Stern et al., 1994; Van Orden et al., 2001). Humans blink every few seconds (Stern et al., 1984),
71 though blinking is less frequent in rodents (Kaminer et al., 2011). Blink rate is modulated by
72 dopaminergic tone (Karson, 1983), and is higher in individuals with schizophrenia (Karson, 1979; Karson
73 et al., 1990). Blinking causes brief decreases in neural activity in visual areas that are similar to transient
74 darkening (Gawne and Martin, 2000; Golan et al., 2016). However, blinking also drives BOLD signals in
75 the visual cortex (Bristow et al., 2005a; Bristow et al., 2005b; Hupe et al., 2012) and somatosensory
76 regions (Guipponi et al., 2015), and blinks are correlated with BOLD activity in the default mode network
77 (Nakano et al., 2013). Blinking is correlated with changes in neural and vascular dynamics across the
78 brain, though the correlates of blinking with arousal are not as well understood.
79 To better understand how pupil diameter and blinking relate to neural activity and cortical
80 hemodynamics across arousal states, we analyzed video of the eye with concurrent monitoring of neural
81 activity and blood volume in the somatosensory cortex of head-fixed mice of both sexes. We investigated
82 the relationship between spontaneous changes in pupil diameter and these physiological signals during the
83 awake state, as well as during REM and NREM sleep. We also tracked how neural activity and cortical
84 blood volume changed around blinking events. Using only changes in pupil diameter and eye position, we
85 found that we can accurately detect and categorize REM and NREM sleep events on a time scale of
86 seconds, providing a simple and robust measure of arousal for studies employing eye monitoring in
87 rodents.
88
89 Materials and Methods
90 Data presented here is from 22 C57BL/6J mice (12 males, Jackson Laboratory, Bar Harbor, ME)
91 between the ages of 3 and 8 months of age. Data from 14 of these mice was previously published (Turner
92 et al., 2020), and we have added imaging/recordings from an additional 8 mice. We obtained a total of
93 442.75 hours of data (20.1 ± 5.3 hours per mouse) of naturally occurring ‘Awake’ (61.4 ± 17.3 %),
94 ‘NREM’ sleep (34.2 ± 16.1%), and ‘REM’ sleep (4.4 ± 2.8%) data.
95 Arousal state nomenclature: Capitalized italics (Rest, NREM, REM, Alert, Asleep, and All) denote
96 arousal states with specific inclusion criteria. Capitalized non-italics with single quotes refer to individual
97 5-second labels of arousal state classified using a machine learning algorithm (‘Awake’, ‘NREM’,
98 ‘REM’) and we use the term ‘Asleep’ (non-italicized) to refer to events classifications as either ‘NREM’
99 or ‘REM’ (Note Fig. 4 and Fig. 5). When generally discussing arousal state in all other contexts, we use
100 the terms awake, sleep, NREM sleep, REM sleep, etc. with no italics or other indicators.
101 Animal procedures. This study was performed in accordance with the recommendations in the Guide
3
102 for the Care and Use of Laboratory Animals of the National Institutes of Health. All procedures were
103 performed in accordance with protocols approved by the Institutional Animal Care and Use Committee
104 (IACUC) of The Pennsylvania State University (Protocol # 201042827). A head-bar, as well as cortical,
105 hippocampal, and nuchal muscle electrodes along with bilateral polished and reinforced thinned-skull
106 windows (Drew et al., 2010; Shih et al., 2012; Zhang et al., 2022b) were surgically implanted under
107 isoflurane anesthesia (5% induction, 2% maintenance). Detailed surgical procedures have been previously
108 described (Mirg et al., 2022a; Mirg et al., 2022b; Turner et al., 2020; Winder et al., 2017). Following
109 surgery, animals were housed individually on a 12-hr. light/dark cycle (lights on at 7:00 am) with food
110 and water ad libitum. Each animal was gradually acclimated to head-fixation in the weeks following
111 recovery. Following the conclusion of imaging experiments, animals were deeply anesthetized and
112 transcardially perfused with heparin-saline followed by 4% paraformaldehyde for histological verification
113 of electrode placement (Adams et al., 2018; Drew and Feldman, 2009).
114 Physiological data acquisition. Data were acquired with a custom LabVIEW (National Instruments,
115 Austin, TX) program (https://github.com/DrewLab/LabVIEW-DAQ). For details on intrinsic optical
116 signal (IOS) imaging, electromyography (EMG), electrophysiology, whisker stimulation, and behavioral
117 measurements see (Turner et al., 2020; Zhang et al., 2022b). Previous work from our lab has shown that
118 the intrinsic signal is not affected by skull/brain movement or other motion artifacts. In previously
119 published experiments (Winder et al., 2017), we looked at reflectance changes in a piece of clay mounted
120 over the cranial window, which would be sensitive to any motion artifacts as well as any other non-
121 hemodynamic noise sources. The reflectance changes were on the order of 0.01%, much smaller than the
122 ~20% changes in reflectance (∆R/R) we see during sleep. Furthermore, 2-photon imaging from our lab of
123 mice running on a treadmill (Gao and Drew, 2016; Echagarruga et al.,. 2020) showed ~2 µm of
124 brain/skull motion during locomotion (a more extreme imaging condition than presented here), which is
125 too small to impact the IOS. IOS reflectance was converted to changes in total hemoglobin (∆[HbT])
126 using the Beer-Lambert law (Ma et al., 2016a; Ma et al., 2016b). Data was acquired in 15-minute
127 intervals with a < 1-minute gap in between for saving data to disk. The vibrissae (left, right, or a third air
128 puffer not directed at the body as an auditory control) were randomly stimulated with air puffs (0.1
129 seconds, 10 pounds-force per square inch (PSI)) occurring every 30–45 second for the first ~1 hr. of
130 imaging.
131 Pupil diameter measurement. The pupil was illuminated with 780 nm light (measured at 0.02–0.05
132 mW/mm2, Thorlabs S120VC) as mice are functionally blind to these wavelengths (Breuninger et al.,
133 2011; Chang et al., 2013). To prevent constriction of the pupil by the illumination for IOS, a dichroic
134 mirror located above the head was used to block as much of the IOS illumination from the eyes as
135 possible so that the mouse was only exposed to a faint glow in its periphery (0.002–0.005 mW/mm2) with
4
136 no visible wavelength light shining directly into its eyes. We did not try to eliminate all green light
137 reaching the eye because green light has sleep-promoting effects (Pilorz et al., 2016). While we did not
138 quantify the pupil diameter in complete darkness as higher baseline dilations would result in the pupil
139 edges being obscured by the eyelid. All pupil diameter measurements were verified by manual inspection
140 (KLT).
141 Data Availability. Data and sample files for running the pupil tracking algorithm are available at
142 https://datadryad.org/stash/share/pv4ZmJnSk65Y6yWxoO6jdb9ou5H-x5wfOJOTCPkBntE and analysis
143 code is available at https://github.com/KL-Turner/Turner-Manuscript2022. Data was analyzed with code
144 written by K.L.T, K.W.G, and P.J.D (MathWorks, MATLAB 2019b-2022a, Natick, MA).
145 Automated pupil diameter measurement and blink detection. Pupil diameter was extracted from
146 videos of the eye taken using a Basler GigE camera (acA640-120gm, Ahrensburg, Germany) with a 75
147 mm double Gauss, fixed focal length lens (#54-691, Edmond Optics Barrington, NJ) at 30 frames/second.
148 Our pupil detection algorithm was adapted from the thresholding in Radon space (TiRS) algorithm
149 developed to determine vessel cross-sections (Gao and Drew, 2014). The sclera and pupil were defined as
150 the area within a user-selected region of interest (ROI) created by outlining the first frame with the eye
151 fully open. Images were inverted so that the pupil had the maximal intensity and were 2-D median filtered
152 ([5,5] x,y pixel median). The pixel intensity distribution was fit with a normal distribution using
153 maximum likelihood estimation (MLE), and pixels above a user-defined threshold (average of 1 ± 0.25
154 standard deviations from the MLE mean) were set to 1 and all other pixel’s intensities were set to 0. The
155 binarized image was then converted to Radon space and normalized within each projection angle (Gao
156 and Drew, 2014). A second threshold was applied prior to conversion back to image space and any holes
157 within the filtered pupil object were filled. Frame-wise pupil area was then calculated using a boundary
158 classification algorithm. Occasional obstructions of the pupil (by a vibrissae) were identified by detecting
159 rapid fluctuations in pupil area. For these frames, the threshold used for thresholding in Radon space was
160 iteratively decreased until the pixel area was within frame-wise change boundaries. Any periods where
161 the pupil was obscured or otherwise unmeasurable were discarded from analysis. The TiRS algorithm was
162 developed to detect small changes in the area of an ellipse (Gao and Drew, 2014). Other techniques,
163 including using a deep neural network (DNN) such as DeepLabCut (Mathis et al., 2018), can be used to
164 track the pupil diameter and position as well (Privitera et al., 2020), though this requires training the
165 network. While the pupil was slightly elliptical in appearance due to the angle of the camera and
166 movement of the eye, the correlation between the major and minor axis of the object’s area was 0.96 ±
167 0.02 (N = 22 mice), indicating the ellipse at the pupil boundary does not appreciably change shape, only
168 size. The area (A) was used to calculate the diameter (d) of the pupil using the formula 2⁄ .
169 MATLAB function(s): roipoly, medfilt2, imcomplement, mle, pdf, radon, iradon, bwboundaries,
5
170 bwconvhull, imfill, regionprops, diff, fillmissing.

171 Blink detection. Blinking was detected independently of pupil diameter from the same ROI as used
172 for pupil diameter measurements. Rapid changes in eyelid motion were detected by using the sum of pixel
173 intensities within the user-defined ROI. The intensity of all pixels for each frame (sclera and pupil) were
174 summed, and a frame-to-frame ROI intensity difference was calculated. As the eyelid obscures the pupil
175 during blinks, rapid changes in overall pixel intensity corresponded to eyelid movements. A threshold
176 was applied to binarize changes in luminance across frames, and blinking events were extracted as above
177 thresholded changes in image intensity. As mice frequently blinked several times in rapid succession,
178 blinking events that occurred with less than 1 second between them were concatenated into a single
179 blinking bout. Periods where a false-positive blink was detected, such as from the eye only partially
180 closing, were also discarded. All detected blinking events were verified by manual inspection (KLT).
181 Electrophysiological analysis. Discrete LFP bands were digitally bandpass filtered from the
182 broadband data using a third-order Butterworth filter into the following bands: delta [1–4 Hz], theta [4–10
183 Hz], alpha [10–13 Hz], beta [13–30 Hz], gamma [30–100 Hz]. The filtered signal was then squared, low-
184 pass filtered < 10 Hz, and resampled at 30 Hz. Time frequency spectrograms were calculated using the
185 Chronux toolbox (Bokil et al., 2010), version 2.12 v03, function mtspecgramc with a 5 second window
186 and 1/5 second step size using [5,9] tapers and a passband of 1–100 Hz to encompass the LFP.
187 Electromyography (EMG, 300 Hz – 3 kHz) from the nuchal (neck) muscles was bandpass filtered,
188 squared, convolved with a Gaussian kernel with 0.5 second standard deviation, log transformed, and then
189 resampled at 30 Hz. MATLAB function(s) butter, zp2sos, filtfilt, gausswin, log10, conv, resample.
190 Sleep scoring. Sleep states were scored consistent with previously published criteria (Cirelli, 2009;
191 Saper and Fuller, 2017; Weber and Dan, 2016). NREM sleep is marked by predominantly elevated
192 cortical delta-band power and lower EMG power during slow-wave (NREM sleep) (Amzica and Steriade,
193 1998; Steriade et al., 1993). REM sleep is marked by elevated hippocampal theta-band power and
194 elevated cortical gamma-band power with even further reduced EMG power (muscle atonia) (Cantero et
195 al., 2004; Le Van Quyen et al., 2010; Montgomery et al., 2008; Sullivan et al., 2014). Periods of user-
196 verified awake rest greater than 5 second in duration with no whisker stimulation, no whisker motion, and
197 no detectable body motion were identified and used baseline characterization of all signals as well as for
198 z-scoring the pupil. Sleep scoring was performed as in (Turner et al., 2020). Every 5 second interval was
199 classified as either Awake, NREM sleep, or REM sleep using a bootstrap aggregating random forest
200 model with the predictors of cortical delta LFP, cortical beta LFP, cortical gamma LFP, hippocampal
201 theta LFP, EMG power, heart rate, and whisking duration. Sleep model accuracy was validated using the
202 out-of-bag error during model training. MATLAB function(s): TreeBagger, oobError, predict.
6
203 Pupil diameter during different arousal states. Pupil diameter was taken from awake resting events
204 (Rest) (≥ 10 seconds in duration), volitional whisking (Whisk) (2–5 seconds), whisker stimulation (Stim)
205 (0.1 seconds, 10 PSI to vibrissa), NREM (≥ 30 seconds), and REM (≥ 60 seconds). Pupil diameter was
206 low-pass filtered < 1 Hz with a fourth-order Butterworth filter. Changes in whisking-evoked and
207 stimulus-evoked (contralateral, auditory) diameters were taken as the change in diameter relative to the
208 mean of the 2 seconds preceding the event onset. Classifications of Alert or Asleep were taken as 15-
209 minute periods with no whisker stimulation and at least 80% of a given classification (Awake for Alert,
210 NREM or REM for Asleep) within a 15-minute recording. The classification of All denotes all data taken
211 during periods with no sensory stimulation, independent of arousal state. MATLAB function(s): butter,
212 zp2sos, filtfilt.
213 Power spectra and coherence. Spectral power was estimated using the Chronux toolbox (Bokil et al.,
214 2010) function mtspectrumc. For pre-whitening spectra, the first derivative was taken of the mean-
215 subtracted data before the power calculation. Gamma-band power measurements were scaled by a factor
216 of 1×1019 so that the magnitude of the neural changes (arbitrary units, a.u.) were more in line with those
217 from the hemodynamic signal for ease of comparison. Coherence analysis was run for each data type
218 using the Chronux function coherencyc. MATLAB function(s): detrend, diff.
219 ∆[HbT]/Gamma-band power vs. pupil relationship. Mean changes in total hemoglobin (∆[HbT]) or
220 gamma-band power (∆P/P), and pupil diameter (z-units) during each arousal state classification (‘Awake’,
221 ‘NREM’, ‘REM’) was plotted as a 2D histogram to highlight clustering in each class. Each classification
222 was assigned a color and the three images were merged as a composite in Fiji (ImageJ). MATLAB
223 function(s): histogram2.
224 Cross-correlation. Data was low-pass filtered < 1 Hz using a fourth-order Butterworth filter and
225 mean-subtracted. Cross-correlation analysis was run for each arousal state with either a ± 5 second lag
226 (Rest, NREM, REM) or ± 30 second lag (Alert, Asleep, All) depending on duration of the behavioral state.
227 MATLAB function(s): xcorr, filtfilt, detrend.
228 Interblink interval and blink-associated physiology. Due to gaps in recording to save the data to disc,
229 interblink interval was calculated between blinks occurring within 15-minute records and not blinks on
230 the edges of trials. Blinks that occurred with 1 second of each other were linked together as blinking
231 bouts, and all blink-triggered analysis were with respect to the first blink in a series. Blink-triggered
232 averages were separated into two groups depending on the arousal state classification of the 5 second bin
233 prior to the blink, that being either Awake (arousal state classification of Awake) or Asleep (arousal state
234 classification being either NREM or REM).
235 Probability of sleep as a function of pupil diameter. The probability of being in each arousal state
236 (‘Awake’, ‘NREM,’ ‘REM’) as a function of pupil diameter was obtained from the mean diameter during
7
237 each 5 second sleep score and binning the value into a histogram, which was then smoothed with a
238 median filter and Savitzky–Golay filter. MATLAB function(s): medfilt1, sgolayfilt.
239 Tracking of pupil location. Motion of the pupil was tracked by using the X and Y coordinates of the
240 centroid obtained during pupil tracking and comparing the change in position to the baseline position
241 during Rest. Frame-by-frame changes in centroid location were low-pass filtered < 10 Hz using a 4th-order
242 Butterworth filter. Motion of the eye was evaluated during each arousal state classification (Awake,
243 NREM, REM) by taking the cumulative sum of the absolute change in centroid position within each 5
244 second bin. Transitions between arousal states (‘Awake’, ‘NREM’, ‘REM’) for pupil diameter, position,
245 and motion were extracted from arousal classifications with 30 seconds of consecutive classifications of
246 one state followed by 30 seconds of another.
247 Eye, physiological, and combined model comparison. The Eye model was trained using only eye
248 metrics, pupil diameter (mean, variance, minimum of both z-unit and mm diameter), position (mean,
249 variance, and maximum displacement of the x,y centroid), and motion (sum, variance of centroid’s
250 absolute velocity) from data with open eyes. The Physiological model was trained using non-eye-based
251 measures (cortical delta power, hippocampal theta power, EMG, etc. - see (Turner et al., 2020)) from the
252 same data so that the training/testing sets were identical among models. The Combined model used a
253 union of all parameters from both models. Each model used was trained using a bagged random forest
254 composed of 128 decision trees, with out-of-bag error and confusion matrices from each animal being
255 calculated at the time of model training using a 70-30 (training-testing) split of randomly shuffled labels
256 taken from each arousal state. MATLAB function(s): TreeBagger, oobError, predict, confusionchart.
257 Experimental design and statistical analysis. Researchers were not blind to the experimental
258 conditions or data analysis. Sample sizes are consistent with those of previously published studies
259 (Echagarruga et al., 2020; Huo et al., 2015; Turner et al., 2020; Winder et al., 2017). Statistical
260 evaluations were made using either generalized linear mixed-effects (GLME) models with the arousal
261 state as a fixed effect, mouse identity as a random effect, and hemisphere (L/R, if applicable) as an
262 interaction with the animal ID; or using a paired t-test where appropriate. Unless otherwise stated,
263 statistical results report p-values from a GLME test. All reported quantifications are mean ± standard
264 deviation unless otherwise indicated. Unless otherwise noted, all pupil diameter measurements are in z-
265 units. MATLAB function(s): fitglme, ttest.
266
267 Results
268 Unanesthetized mice were head-fixed under an IOS imaging setup (Huo et al., 2014; Pisauro et al.,
269 2013; Sirotin and Das, 2009; Vazquez et al., 2014; Winder et al., 2017) with concurrent electrophysiology
270 to measure changes in neural activity from the vibrissa region of the somatosensory cortex (Petersen,
8
271 2007; 2014) and hippocampal CA1. We also tracked whisker motion (Winder et al., 2017),
272 electromyography of the nuchal muscles (Turner et al., 2020), and pupil diameter/blinking (Larsen and
273 Waters, 2018; Reimer et al., 2014; Reimer et al., 2016; Vinck et al., 2015) (Fig. 1a). Using the
274 thresholding in Radon space algorithm (Gao and Drew, 2014), we detected the outline of pupil from video
275 frames (Fig. 1b-e, Video 1) in order to quantify diameter changes. Blinks were detected from rapid pupil
276 diameter changes (Fig. 1f). During the awake state, there are frequent bouts of whisker motion that are
277 correlated with increases in pupil diameter, and the EMG has high power. The LFP has reduced low-
278 frequency power and hemodynamic fluctuations have low amplitude, except during periods of extended
279 behavior or sensory stimulation. During NREM sleep, whisker motion and EMG power are much lower
280 and the pupil diameter is smaller than in the awake state. Cortical LFP and hemodynamic signals both
281 begin to increase in amplitude with low frequency oscillations in broadband LFP power. During REM
282 sleep, whisker motion increases along with eye movement, but the pupil remains constricted. Power in the
283 EMG is at its lowest point due to muscle atonia (with occasional twitches), blood volume increases
284 substantially above both awake and NREM levels (Bergel et al., 2018), and a prominent theta-band is
285 visible in the hippocampal LFP. A representative example of how pupil diameter changes along with
286 fluctuations in ∆[HbT], hippocampal LFP, and other behavioral cues during each arousal state is shown in
287 Fig. 1f and Video 1. Since mice have been shown to sleep with their eyes open during both head-fixed
288 (Karimi Abadchi et al., 2020; Turner et al., 2020; Yuzgec et al., 2018) and freely moving (Senzai and
289 Scanziani, 2022) conditions, we tested whether neural activity and hemodynamics might differ between
290 the eyes-open and eyes-closed instances of the REM state. We compared the eyes-open REM sleep
291 physiology with that during obtained during a few instances of REM sleep with eyes-closed in a subset of
292 mice (N = 8) that was excluded from subsequent analyses due to the inability to measure pupil diameter.
293 The change in total hemoglobin (∆[HbT]) during REM with eyes open (73.9 ± 12.9 µM) was not
294 significantly different from the hemoglobin changes during REM sleep with eyes closed (76.2 ± 21.2 µM,
295 p < 0.49, paired t-test). The difference in theta-band power in the hippocampus during eyes-open REM
296 and eyes-closed REM was also not statistically significant, (1.06 ± 1.11 A.U. vs. 1.06 ± 1.10 A.U.,
297 respectively, p < 0.96, paired t-test). The lack of detectable difference between the eyes-open and eyes-
298 closed sleep state suggests that whether or not they eyes are open has little impact on sleep physiology.
299
300 Quantification of pupil diameter across arousal states
301 To quantify how fluctuations in pupil diameter change with arousal state, we compared the pupil’s
302 diameter during several distinct arousal states and behaviors. Pupil diameter was largest during periods of
303 arousal and smallest during sleep. To standardize measures across animals and to account for slight
304 differences in baseline illumination intensity, we z-scored the pupil diameter to the mean and standard
9
305 deviation during all periods of awake quiescence lasting at least 5 seconds in length. Italics denote periods
306 meeting our arousal state criteria. 15-minute periods with at least 80% of its model scores as ‘Awake’
307 were classified as Alert. 15-minute periods with 80% of the time in ‘NREM’ or ‘REM’ sleep were
308 classified as Asleep. All was all data, regardless of arousal state. Periods with whisker stimulation were
309 excluded from all states. Two mice did not have any 15-minute periods that met the criteria for the Alert
310 or Asleep categories (N = 20 mice included) with the other four behaviors (Rest, NREM, REM, All) all
311 having N = 22 mice. Rest, defined as all periods of awake quiescence lasting at least 10 seconds in length,
312 had a mean pupil diameter of 0.6 ± 0.17 mm (Fig. 2a) or -0.25 ± 0.88 z-units (Fig. 2b). We used a GLME
313 model to compare the pupil’s diameter among states, using the Bonferroni correction for multiple
314 comparisons (10), which puts the adjusted significance threshold (α) at 0.005. The mean z-unit during
315 Rest was non-zero because the minimum duration used for z-scoring (5 seconds) was shorter than the
316 minimum duration for inclusion in Rest (10 seconds). During volitional whisking bouts lasting 2–5
317 seconds, pupil diameter increased to 0.89 ± 0.17 mm (p < 2.3×10-19 versus Rest) or 3.39 ± 0.88 z-units (p
318 < 2.1×10-22). Following brief stimulation of the vibrissae, pupil diameter increased to 0.75 ± 0.18 mm (p <
319 9.2×10-8) or 1.95 ± 1.82 z-units (p < 1.9×10-11). The pupil diameter decreased during NREM (> 30
320 seconds in length) to 0.35 ± 0.07 mm (p < 8.1×10-16) or -3.43 ± 0.64 z-units (p < 6.2×10-19) and even
321 further during REM (> 60 seconds in length) to 0.26 ± 0.03 mm (p < 3.2×10-23) or -4.56 ± 0.75 z-units (p
322 < 2.3×10-27). Additional statistical comparisons for the difference in pupil size between each arousal state
323 can be found in Table 1 and Table 2. Periods of volitional whisking and whisker stimulation caused
324 increases in pupil diameter of 1.44 ± 0.47 z-units and 1.61 ± 0.62 z-units, respectively. Auditory controls
325 dilated the pupil by 0.98 ± 0.58 z-units (Fig. 2c). Note that while the z-unit diameter was set relative to
326 awake resting events, these increases are with respect to changes in z-unit diameter relative to the 2
327 seconds prior to event onset, which was a mixture of awake resting and volitional behaving data of
328 various durations.
329
330 Table 1 | Statistical comparison of Mean pupil diameter across arousal states (mm)
Whisk Stim NREM REM
Rest ***2.3×10-19 ***9.2×10-8 ***8.1×10-16 ***3.2×10-23
-7
Whisk ***5.4×10 ***1.2×10-38 ***2.8×10-44
Stim ***2.3×10-28 ***6.0×10-35
NREM *0.0012
331 Bonferroni corrected significance levels (10): *α < 0.005, **α < 0.001, ***α < 0.0001
332
333 Table 2 | Statistical comparison of Mean pupil diameter across arousal states (z-units)
Whisk Stim NREM REM
Rest ***2.1×10-22 ***1.9×10-11 ***6.2×10-19 ***2.3×10-27
-6
Whisk ***3.1×10 ***2.3×10-43 ***2.2×10-49
Stim ***1.2×10-34 ***1.3×10-41
NREM **0.00019
10
335
336 To look at changes in the power spectrum of the pupil diameter during different arousal states, we
337 pre-whitened the pupil diameter throughout analysis during periods of continuous Rest (> 10 second
338 duration), continuous NREM (> 30 second duration), continuous REM (> 60 second duration), and 3
339 additional longer-length classifications (Alert, Asleep, and All) by taking the first temporal derivative of
340 the z-unit diameter prior to power estimation. Except for REM, the pre-whitened power spectra of the
341 pupil diameter during the longer length behaviors were similar to each other at the lower frequencies (Fig.
342 2d). These findings support previous reports that the pupil diameter increases with arousal and attention
343 but decreases during periods of sleep (Karimi Abadchi et al., 2020; Yuzgec et al., 2018).
344
345 Relationship between pupil diameter and hemodynamic and neural signals
346 We next asked how well the pupil diameter correlated with both hemodynamic and neural signals in
347 the somatosensory cortex of mice both during awake behaviors and during different sleep states. For each
348 mouse, the pupil diameter was compared to both hemispheres’ neural and hemodynamic signals, yielding
349 2*N measurements for each behavior. The non-independence (due to within animal correlations) was
350 accounted for as an interaction term in the statistical comparison (see Materials and Methods). During the
351 ‘Awake’ state, changes in pupil diameter and gamma-band power were small. When the animals were in
352 ‘REM’ and ‘NREM’ sleep, the pupil constricted and gamma-band power increased substantially relative
353 to the ‘Awake’ state (Fig. 3a). We then looked at the coherence and the cross-correlation between the
354 envelope of gamma-band power (∆P/P) and pupil diameter. The minimum cross-correlation during Rest
355 was -0.16 ± 0.20 at a lag of -0.07 seconds (Fig. 3b) and during Alert was -0.02 ± 0.14 at 0.1 seconds (Fig.
356 3c), indicating very little time difference between cortical gamma-band and corresponding pupil diameter
357 fluctuations during awake behaviors. Extrema in the cross-correlation in NREM were: -0.26 ± 0.06 at -0.1
358 seconds; REM: 0.05 ± 0.04 at 1.53 seconds; Asleep: -0.32 ± 0.05 at -0.43 seconds; and All: -0.23 ± 0.11 at
359 -0.2 seconds.
360 We next evaluated the coherence between gamma-band power and pupil diameter during different
361 arousal states (using GLME models) and at two different frequencies. The specific frequencies (0.02, 0.35
362 Hz) were chosen for statistical comparison as they were the approximate peaks in pupil-gamma power
363 coherence during the Asleep and Alert behavioral states. Using the Bonferroni correction for multiple
364 comparisons (3 at 0.02 Hz, 15 at 0.35 Hz) put the adjusted significance thresholds (α) at 0.017 and 0.003
365 respectively. The pupil-gamma power coherence during Alert periods at 0.02 Hz was 0.34 ± 0.18 (Fig.
366 3d, e, Table 3), and was significantly elevated during periods of Asleep at 0.68 ± 0.09 (p < 2.4×10-18) and
367 during All data at 0.48 ± 0.19 (p < 2.9×10-5). Pupil-gamma power coherence at 0.35 Hz (Fig. 3d, f, Table
368 4) during periods of awake Rest and Alert were similar, at 0.2 ± 0.1 and 0.2 ± 0.09, respectively (p <
11
369 0.58), similar to what is seen between the pupil diameter and cortical neuron membrane potential
370 (McGinley et al., 2015). The coherence of all other arousal states at 0.35 Hz were significantly lower than
371 those of the non-sleep states; NREM: 0.09 ± 0.06 (p < 2.9×10-13); REM: 0.14 ± 0.08 (p < 3.7×10-5);
372 Asleep: 0.11 ± 0.07 (p < 9.8×10-9); All: 0.11 ± 0.08 (p < 5.3×10-9).
373
374 Table 3 | Statistical comparisons of pupil-gamma power coherence at 0.02 Hz
Asleep All
Alert ***2.4×10-18 **2.9×10-4
Asleep ***5.0×10-9
376
377
378 Table 4 | Statistical comparisons of pupil-gamma power coherence at 0.35 Hz
NREM REM Alert Asleep All
Rest ***2.9×10-13 ***3.7×10-5 0.58 ***9.8×10-9 ***5.3×10-9
NREM **0.0005 ***3.2×10-14 0.12 0.10
REM ***5.6×10-6 0.07 0.07
Alert ***1.2×10-9 ***6.2×10-10
Asleep 0.96
380
381 We then looked at how the relationship between pupil diameter and cortical blood volume changed
382 with arousal state. During the ‘Awake’ state in the absence of stimulation, fluctuations in pupil diameter
383 and hemodynamic signals were small. When the animals were in ‘REM’ and ‘NREM’ sleep, the pupil
384 constricted, while blood volume increased substantially (Fig. 3g). We then looked at the cross-correlation
385 between ∆[HbT] and pupil diameter to determine their temporal relationship. We saw the extrema of the
386 cross-correlation during awake Rest was -0.37 ± 0.25 at a lag of 1.13 seconds, such that pupil diameter
387 changes occurred on average 1.13 seconds before hemodynamic oscillations, and pupil diameter was
388 negatively correlated with blood volume (Fig. 3h). NREM was more strongly anti-correlated than Rest at -
389 0.56 ± 0.08 at 0.73 seconds. REM had a maximum positive correlation of 0.24 ± 0.11 at 1.97 seconds. The
390 peak of the ∆[HbT] and pupil diameter correlation in the Alert condition was 0.1 ± 0.21 at -0.43 seconds;
391 this negative lag is likely due to sustained bouts of whisking (Fig. 3i). Asleep and All were anti-correlated
392 at -0.65 ± 0.08 at a lag time of 1.0 seconds, and -0.48 ± 0.15 at a lag of 1.23 seconds respectively.
393 Generally, pupil dilations preceded vasoconstriction by about a second.
394
395 Table 5 | Statistical comparisons of pupil-∆[HbT] coherence at 0.02 Hz
Asleep All
Alert ***2.0×10-29 ***9.3×10-14
Asleep ***1.3×10-10
397
12
398 Table 6 | Statistical comparisons of pupil-∆[HbT] coherence at 0.35 Hz

NREM REM Alert Asleep All
Rest ***7.0×10-33 ***9.8×10-66 0.33 ***2.5×10-33 ***3.4×10-8
NREM ***2.6×10-19 ***6.9×10-35 0.59 ***1.4×10-14
REM ***1.0×10-66 ***7.7×10-17 ***7.1×10-47
Alert ***2.6×10-35 ***3.9×10-10
Asleep ***1.9×10-15
400
401 For nearly all cases, the coherence between hemodynamic signals and pupil diameter was
402 substantially higher at lower frequencies. We evaluated the pupil-∆[HbT] coherence during different
403 arousal states at 0.02 Hz (Fig. 3j, k, Table 5) and 0.35 Hz (Fig 3j, l, Table 6). As with the comparisons
404 between pupil diameter and gamma-band power, these specific frequencies were chosen for statistical
405 comparison (GLME) as they were the approximate peaks in pupil-∆[HbT] coherence during Asleep
406 behaviors and Alert behaviors, respectively. A Bonferroni correction for multiple comparisons (3
407 comparisons at 0.02 Hz, 15 at 0.35 Hz) put the adjusted significance threshold (α) at 0.017 and 0.003,
408 respectively. At 0.02 Hz, coherence between total hemoglobin and pupil diameter during the Alert state
409 was 0.32 ± 0.18 (N = 20 mice). Since we can only evaluate the 0.02 Hz component of the spectrum in the
410 longer 15-minute duration epochs, statistical comparisons only done for longer duration behavioral
411 conditions. Pupil-∆[HbT] coherence during Asleep periods, was 0.79 ± 0.06 (p < 2.0×10-29) and All
412 periods was 0.58 ± 0.17 (p < 9.3×10-14). At 0.35 Hz, Rest had a pupil-∆[HbT] coherence of 0.51 ± 0.06 (N
413 = 22 mice). The pupil-∆[HbT] coherence during rest was not significantly different than that during the
414 Alert behavior, being 0.52 ± 0.09 (p < 0.33). However, the coherence at 0.35 Hz was significantly lower
415 during all sleep associated periods – NREM: 0.3 ± 0.11 (p < 7.0×10-33); REM: 0.15 ± 0.08 (p < 9.8×10-66);
416 Asleep: 0.29 ± 0.12 (p < 2.5×10-33); All: 0.42 ± 0.1 (p < 3.4×10-8).
417 Low-frequency fluctuations in blood volume are highly correlated with the pupil diameter, suggesting
418 that the same processes that modulate arousal on these timescales also cause constriction in the cerebral
419 vasculature. There was a strong negative correlation between pupil diameter and blood volume as well as
420 between pupil diameter and gamma-band power during periods of Rest and Asleep, with these
421 correlations turning positive during periods of alertness and active whisking. This is in general consistent
422 with an anticorrelation between arousal level and blood volume (Cardoso et al., 2019) in the absence of
423 stimulation or active whisking behavior.
424
425 Blinking is associated with increases in arousal
426 We next explored the relationship between blinking and arousal level. The mean inter-blink interval
427 was 117.4 ± 33.6 seconds (Fig. 4a, b, N = 22 mice). We then quantified the probability of a blink being
428 elicited by sensory stimulation of the whiskers. Except for one animal, our mice did not frequently blink
13
429 after a whisker stimulation, with the mean probability of blinking within 5 seconds post-stimulus being
430 only 5.4 ± 5.1% (Fig. 4c, N = 21 mice, excluded mouse at 25.6%). For blinks that did occur following
431 stimulation, blinks typically occurred within a half-second (~37%) and the probability decreased
432 thereafter. Blinks primarily occurred during the ‘Awake’ state (> 70%), but when they did occur during
433 periods of ‘REM’ or ‘NREM’ sleep the animal either quickly returned to sleep after a brief awakening or
434 did not wake up at all (Fig. 4d). Blinking events also occurred during periods of maintained REM sleep
435 with no awakening. The likelihood of volitional whisking occurring simultaneously with blinking
436 behavior was high (Fig. 4e), with volitional whisking occurring during 40% of ‘Awake’ blink events.
437 Whisking was less frequent during blinks that occurred either during or immediately following periods
438 classified as ‘Asleep’ (with the majority being in ‘NREM’).
439 As blinks commonly occurred in clusters, for all blink-related analysis we only looked at the first
440 blink occurring in a string of blink events that occurred within less than 1 second of each other. All blinks
441 that occurred within ± 5 seconds of whisker stimulation were also excluded to prevent the stimulus from
442 interfering with the blink-triggered comparison. To help separate the effect of the blink from the effects`
443 of volitional whisking behavior coinciding with it, we split the data set into low-whisk blinks (LWB) and
444 high-whisk blinks (HWB), further split by whether they occurred when the mouse was ‘Awake’ or
445 ‘Asleep’ (‘REM’ and ‘NREM’ combined). LWB were defined as events where the mouse was whisking
446 for less than 1/3 of a second in the 2 seconds surrounding each blink, whereas HWB were defined as
447 whisking for greater than 1 second within ± 2 seconds of the blink. We evaluated blink-triggered averages
448 for changes in diameter (z-units), ∆[HbT], EMG power, cortical LFP, and hippocampal LFP across all
449 blinks from all mice that met the criteria. During the ‘Awake’ state, LWB caused an increase of around
450 0.7 z-units in pupil diameter compared to resting baseline, while HWB caused an increase in excess of 3
451 z-units (Fig. 4f). ∆[HbT] during LWB increased around 4 µM from the period prior to the blink, whereas
452 HWB increased around 10 µM, probably due to the increased whisking in the time preceding the blink
453 (Fig. 4g). EMG power increases ~0.6 orders of magnitude during LWB compared to 1.2 orders of
454 magnitude during HWB, with power increases starting about 0.5 seconds before the blink occurs. There
455 were negligible increases in gamma-band power in both the somatosensory cortex (Fig. 4i) and the
456 hippocampus (Fig. 4k) during LWB, with the ~20% increases in power seen during high whisking likely
457 being due to the whisking (Winder et al., 2017) and not the blink (Fig. 4j, l).
458 Blink-associated neural and vascular changes were similar during blinks that occurred during periods
459 of ‘Asleep’, with the caveat that the baseline pupil diameter was lower. Pupil diameter during both high-
460 and low-whisk asleep blinks was small prior to the blink (< -3 z-units), with the diameter slowly
461 increasing preceding the blink as the animal began to wake up. For periods of low whisking, the pupil
462 remained constricted as the animals presumably fell back asleep or remained asleep (Fig. 4m). Changes
14
463 in hemodynamics followed a similar trend, with there being a reduction in total hemoglobin in the time
464 preceding the blink as the animals are waking up (Fig. 4n). EMG power changes were similar to the
465 ‘Awake’ blink, but with a lower baseline power (Fig. 4o). We noted no appreciable differences among the
466 blink-associated power spectra during blinks that occurred around sleeping periods (Fig. 4p, q, r, s),
467 however the drop in cortical delta power around the blink is evident as the animals wake up from sleep
468 (which was predominantly NREM). Altogether, blinking was associated with increases in arousal, but to
469 varying degrees marked by differing amounts of accompanying whisking. When blinks occurred during
470 the ‘Awake’ state, there was only a small increase in blood volume if there was minimal whisking.
471 However, when blinks occurred during the ‘Asleep’ state, large decreases in blood volume followed
472 regardless of whisking amount.
473 Awake blinking causes a resetting of neural and vascular dynamics
474 We next asked how blinking is correlated with the neural and vascular synchrony between bilateral
475 regions of somatosensory cortex. Processing of new or startling information is accompanied by blinking,
476 suggesting it is associated with some sort of mental ‘resetting’ (Siegle et al., 2008). As before, we
477 separated blinks by arousal state (‘Awake’, ‘Asleep’) and looked at the changes in power and coherence
478 during periods preceding (-15 seconds to -5 seconds) and following (5 seconds to 15 seconds) each blink.
479 We excluded the time ± 5 seconds adjacent to each blink to capture the pre- and post- effects of the blink
480 on synchrony and not the blink itself, as the blink elicits brief changes in neural activity and blood
481 volume, and this response will drive coherence across all frequencies temporally close to the blink that
482 reflect the evoked activity, not necessarily a state change in the brain. For both bilateral cortical gamma-
483 band power signals and bilateral changes in ∆[HbT], we evaluated the power/coherence changes from 0 to
484 0.2 Hz as this corresponded to the frequency domain of the largest difference between pre- and post-
485 blink. One animal was excluded from the analysis in Fig. 4 due to being an outlier in its blink rate (N =
486 21). Coherence between bilateral gamma-band power signals during the ‘Awake’ state dropped following
487 a blink from 0.38 ± 0.13 (pre-blink) to 0.28 ± 0.10 (post-blink) (p < 0.001, paired t-test) (Fig. 5a, b).
488 Power in these signals (Fig. 5c) did not change in the period following a blink, 13.2 ± 81.0 a.u. (pre-
489 blink) versus 11.9 ± 73.7 a.u. (post-blink) (p < 0.25, paired t-test) (Fig. 5d). There was no clear
490 relationship between changes in gamma-band power pre- vs. post-blink vs. changes in bilateral gamma-
491 band coherence pre- vs. post-blink (Fig. 5e). ‘Awake’ bilateral ∆[HbT] coherence was 0.88 ± 0.03 (pre-
492 blink) dropping to 0.84 ± 0.05 (post-blink) (p < 6.4×10-6) (Fig. 5f, g). Unlike the gamma-band power,
493 there was a significant drop in cortical hemodynamic power following a blink: 149.1 ± 49.1 a.u. (pre-
494 blink) vs. 100.4 ± 33.3 a.u. (post-blink) (p < 8.2×10-13, paired t-test) (Fig. 5h, i), and these appear to be
495 related (Fig. 5j).
15
496 For blinks that occurred during the ‘Asleep’ state, neither the gamma-band power nor total
497 hemoglobin showed any significant changes between pre- and post-blink power or coherence. The mean
498 coherence from 0-0.2 Hz between bilateral gamma-band signals during ‘Asleep’ blinks was 0.44 ± 0.18
499 (pre-blink) versus 0.46 ± 0.17 (post-blink) (p < 0.38, paired t-test) (Fig. 5k, l) with power changes of 7.0
500 ± 34.4 (pre-blink) versus 7.1 ± 39.9 (post-blink) (p < 0.98, paired t-test) (Fig. 5m, n) with no relationship
501 between power and coherence changes (Fig. 5o). Bilateral hemodynamic coherence during ‘Asleep’
502 blinks was 0.89 ± 0.04 (pre-blink) versus 0.89 ± 0.04 (post-blink) (p < 0.45, paired t-test) (Fig. 5p, q) at a
503 power of 187.1 ± 67.1 (pre-blink) versus 174.1 ± 47.3 (post-blink) (p < 0.17, paired t-test) (Fig. 5r, s)
504 with no relationship between power and coherence changes (Fig. 5t). These results indicate that blinking
505 during the ‘Awake’ state was correlated with a ‘resetting’ of neural and vascular signals, but this did not
506 happen with blinks during ‘Asleep’.
507
508 Pupil size, position, and motion change with arousal state
509 When looking at the probability of being in a given arousal state as a function of pupil diameter (Fig.
510 6a), it is apparent that the probability of wakefulness decreases dramatically as the pupil decreases in size
511 with the 50% probability (that is, equal probability of being awake or asleep) being around -2 z-units from
512 the resting baseline. However, the pupil size in our data set was similar across REM and NREM sleep
513 states. We next asked how other eye metrics could help differentiate further between REM and NREM
514 sleep, as previous work has found systematic changes in eye position of rodents during REM (Sanchez-
515 Lopez and Escudero, 2011). When the animals fall asleep, the pupil begins to drift as the muscles around
516 the eye relax. There are also ‘rapid-eye-movements’ characteristically seen during REM sleep. We
517 tracked the centroid of the pupil (Fig. 6b) and changes in its position measured relative to the resting
518 location. When looking at the eye velocity in each arousal state (Fig. 6c), we see that the pupil moves
519 significantly more during ‘REM’ sleep (0.66 ± 0.41 mm/sec) than in either the ‘Awake’ (0.19 ± 0.05
520 mm/sec, p < 1.8×10-5) or ‘NREM’ states (0.22 ± 0.15 mm/sec, p < 1.5×10-5). Despite a large change in
521 diameter, there was minimal change in pupil centroid location during transitions between the ‘Awake’
522 and ‘NREM’ states (Fig. 6d, e). However, the centroid of the pupil moves between 0.1 and 0.2 mm in the
523 nasal and ventral directions when the animal transitions into the ‘REM’ state (Fig. 6f) and quickly returns
524 to the baseline location upon awakening (Fig. 6g).
525
526 Eye metrics are an accurate predictor of arousal state
527 Finally, we explored using eye metrics (pupil diameter and position) alone as a predictor of arousal
528 state, comparing their predictive power to that of more conventional sleep scoring using physiological
16
529 parameters (cortical delta power, hippocampal theta power, EMG). As pupil diameter is commonly
530 measured in behaving mouse paradigms (International Brain et al., 2021; McGinley et al., 2015; Musall et
531 al., 2019; Pisauro et al., 2016; Stringer et al., 2019; Vinck et al., 2015), arousal scoring using pupil
532 diameter and eye movement could be very useful for detecting bouts of sleep when other more invasive
533 assays are not available. Previous work has shown that the sleep/wake state of head-fixed mice can be
534 determined accurately on the timescale of hundreds of seconds using pupil diameter (Yuzgec et al., 2018),
535 but here we asked if it can be done on a time scale of seconds.
536 Pupil diameter alone could be used to differentiate the ‘Awake’ arousal state from the ‘REM’ and
537 ‘NREM’ states (with a threshold at approximately -2 z-units, Fig. 6a). However, there was much less
538 difference in pupil diameter between the two sleep states. Therefore, we used the position and motion of
539 the pupil’s centroid in our model classification to achieve a separation between ‘REM’ and ‘NREM’. We
540 compared the manual scores of an exemplar 15-minute imaging session to those produced by three
541 different bootstrapped random forest classification models (Fig 7a) following model training using a
542 70:30 training:testing split. The first model utilized only eye metrics, such as the pupil’s diameter (mm,
543 z), position, and average velocity (‘Eye Model’) (see Materials and Methods). We used the same example
544 data shown in Fig. 1f to show how these three metrics can be used to identify arousal state (Fig. 7b, c, d).
545 The pupil decreases in size during ‘REM’ and ‘NREM’ sleep as compared to the awake state, but it is the
546 changes in pupil position that separate ‘REM’ from ‘NREM’ sleep. The second model was identical to
547 that used in (Turner et al., 2020) and used seven physiological parameters (i.e., ‘Physiological Model’,
548 see Materials and Methods). The third model was a combination of the two (‘Combined Model’), utilizing
549 both eye metrics and the physiological measurements together. With respective to identifying sleep, the
550 Eye Model (N = 22 mice) had a total cumulative accuracy of 90.4% across all testing data (Fig. 7e) with
551 comparable type-I and type-II classification errors. The Physiological Model (cumulative accuracy of
552 93.0%) (Fig. 7f) and combined model (cumulative accuracy of 93.9%) (Fig. 7g) performed better than the
553 eye model, having access to the “gold-standard” of electrophysiology and electromyography information
554 classically used for detecting sleep. The out-of-bag error during model training was higher for the Eye
555 Model (0.10 ± 0.02) than the Physiological Model (0.08 ± 0.02, p < 7.0×10-6), as well for as the
556 Combined Model (0.07 ± 0.02, p < 5.2×10-10) (Fig. 7h). However, the eye metrics did significantly
557 improve the Combined Model over the original Physiological Model (p < 1.0×10-6). Our findings indicate
558 that monitoring the eye, including pupil diameter and its changes in position, can be used as a non-
559 invasive method to determine if head fixed mice are sleeping on a timescale of seconds.
560
561 Discussion
17
562 In this study, we explored the relationship between pupil diameter and blinking with ongoing neural
563 activity and hemodynamic signals in the somatosensory cortex of mice during different arousal and
564 behavioral states. Pupil diameter was consistently smaller during sleep states and larger during the awake
565 state, allowing the pupil to be used as a predictor of arousal state (Yuzgec et al., 2018). As a practical
566 matter, a threshold of approximately 2 z-units below the resting pupil diameter can function as an
567 indicator of sleep. Blinking was correlated with changes in arousal levels, as well as with neural and
568 vascular dynamics. Mice were more likely to be in the awake state after blinking than before, and bilateral
569 synchronization in the gamma-band envelope and accompanying hemodynamics both decreased after
570 blinks when the mouse was awake. The difference in neural responses to blinks in the awake versus
571 asleep conditions could be due to differences in sensory processing in these respective states.
572 Because our mice slept frequently, when averaged over the entire data set the pupil diameter was
573 strongly anti-correlated with both gamma-band power and blood volume due to the large vasodilation
574 occurring during NREM sleep (Turner et al., 2020). When restricted to just data in the awake state (whose
575 neural and hemodynamic signals are dominated by body movements and fidgeting behavior (Drew et al.,
576 2020; Drew et al., 2019; Huo et al., 2014; Musall et al., 2019; Salkoff et al., 2020; Stringer et al., 2019;
577 Tran et al., 2018; Winder et al., 2017), the pupil diameter was positively correlated with gamma-band
578 power and blood volume (Fig. 8). The coherence between blood volume and pupil diameter was stronger
579 at lower frequencies and was highest (> 0.75) at 0.02 Hz (corresponding to a ~50 second period) in the
580 sleeping mouse. Interestingly, activity in the LC and the concentration of noradrenaline during NREM
581 sleep fluctuates on a similar timescale (Kjaerby et al., 2022; Osorio-Forero et al., 2021). As noradrenaline
582 is vasoconstrictory (Bekar et al., 2012; Goadsby et al., 1985; Raichle et al., 1975), the fluctuations in
583 noradrenaline during NREM may contribute to the oscillations in blood volume (Fultz et al., 2019; Turner
584 et al., 2020) that simulations and experiments have suggested are important for clearing waste from the
585 brain (Kedarasetti et al., 2020a; 2022; Kedarasetti et al., 2020b; van Veluw et al., 2020; Xie et al., 2013).
586 The fact that there is vasodilation in somatosensory cortex during periods of body motion and arousal
587 when noradrenaline levels are highest can be accounted for by the action of local vasodilatory signals
588 from neurons during behavior (Echagarruga et al., 2020; Winder et al., 2017; Zhang et al., 2019). While
589 our LFP measurements were conducted in the somatosensory cortex, we would expect similar dynamics
590 in other sensory areas of the cortex, such as visual areas, which are innervated by the same LC neurons
591 (Kim et al., 2016). The vascular dynamics may differ in the frontal cortex and other areas, which receive
592 input from a different subset of LC neurons (Kim et al., 2016).
593 Our findings that the pupil diameter was strongly anticorrelated with blood volume should be taken in
594 the context of the literature relating arousal and pupil diameter to LC activity, and the vasoconstrictory
595 impact noradrenergic inputs of LC on the cerebral vasculature. The noradrenaline levels in the brain
18
596 correspond with LC neuron cell body activity (Berridge and Abercrombie, 1999; Feng et al., 2019; Poe et
597 al., 2020). Sensory stimuli drive increases in LC neural activity, noradrenergic tone, and pupil dilation
598 (Gilzenrat et al., 2010; Gray et al., 2021; Joshi et al., 2016; Murphy et al., 2014; Schwarz and Luo, 2015;
599 Yang et al., 2021) though the correlations between pupil diameter and spiking of individual LC neurons is
600 low (Megemont et al., 2022). Activation of noradrenaline-releasing neurons in the LC drive awakening
601 from sleep (Carter et al., 2010; Hayat et al., 2020; Kjaerby et al., 2022), and this awakening from sleep is
602 followed by profound cortical vasoconstriction (Turner et al., 2020) and brain-wide hemodynamic
603 changes (Fultz et al., 2019; Liu et al., 2018; Liu et al., 2015; Zhang et al., 2022a). Electrical stimulation of
604 the LC and subsequent increases in noradrenaline levels (Bekar et al., 2012) cause vasoconstriction
605 (Goadsby et al., 1985; Raichle et al., 1975) (but see (Toussay et al., 2013) that observe that stimulation of
606 the LC can increase cortical perfusion in the anesthetized rat). We saw that the blood volume-pupil
607 diameter correlation was positive in the Alert state, which seems to contradict the vasoconstrictory nature
608 of noradrenaline. The Alert state contains many fidgeting and whisking bouts, which can be nearly
609 continuous body movements. While cortical noradrenaline levels are known to rise during locomotion
610 (Paukert et al., 2014; Polack et al., 2013) and movement (Feng et al., 2019), the activity of vasodilatory
611 neuronal nitric oxide synthase (nNOS) neurons (Echagarruga et al., 2020) and vasodilatory extracellular
612 potassium are both elevated during locomotion (Longden et al., 2017; Rasmussen et al., 2019), and these
613 two vasodilators are likely strong enough to drive vasodilation even the face of elevated noradrenaline.
614 Furthermore, the magnitude of noradrenaline increases during movement are much smaller than the
615 decreases during sleep. During sleep, microdialysis measured cortical noradrenergic levels fall > 50%
616 (Bellesi et al., 2016), though recent studies using fluorescent biosensors have shown there are large
617 fluctuations over the timescales of minutes in noradrenaline levels in frontal areas during NREM (Kjaerby
618 et al., 2022). The arousal-induced noradrenergic levels increases during voluntary body movements and
619 fidgeting are of a substantially smaller magnitude than the decreases seen during sleep (Feng et al., 2019),
620 which would explain why vasodilation dominates in these behaviors. The positive correlations between
621 pupil diameter and blood volume during REM might be due to fluctuations in cholinergic drive, which is
622 high during REM sleep (Jing et al., 2020; Jones, 2020) and is also linked to pupil dilations (Larsen and
623 Waters, 2018). Our observations of state-dependent correlation between blood volume and pupil diameter
624 (Fig. 8) is consistent with noradrenergic modulation in the cortex playing a state-dependent role in
625 neurovascular coupling in concert with local neural activity (Hamel, 2006; Kleinfeld et al., 2011).
626 Monitoring the pupil (Privitera et al., 2020) can provide a second-by-second insight into the
627 sleep/wake state of mice. These periods of sleep can be an issue not only in behavioral tasks but can also
628 be a confound in studies looking at spontaneous neural and hemodynamic activity. In human studies,
629 sleep episodes occur frequently during resting-state imaging (Tagliazucchi and Laufs, 2014), drastically
19
630 affecting functional connectivity measurements and other measures of network dynamics. In mice,
631 bilateral neural and hemodynamic correlations are much higher during sleep (Turner et al., 2020),
632 supporting the need for arousal-state monitoring in both animal and human studies as detecting and
633 monitoring changes in arousal is essential for accurate interpretations of any resting-state study (Drew et
634 al., 2020; Liu et al., 2018; Tagliazucchi and Laufs, 2014). Fortunately, monitoring pupil diameter and eye
635 motion can provide a simple, non-invasive way of detecting sleep and monitoring arousal state transitions
636 in rodents.
637 Fig. 1 | Pupil diameter tracks arousal state
20
638 (a) Photograph and schematic of the experimental setup. IOS imaging of the whisker portion of
639 somatosensory cortex collected using illumination at an isosbestic point of hemoglobin (530 nm or 568
640 nm) through bilateral thinned-skull windows. Electrodes were implanted into layer 5 of the
641 somatosensory cortex and into hippocampal CA1. EMG electrodes were also implanted into the nuchal
642 muscles of the neck. A camera is directed towards the whiskers to track their movement. Another camera
643 captures the eye, which is illuminated by infrared 780 nm light, to observe pupil diameter and blinking.
644 Directed air puffers allow controlled stimulation of the left and right whiskers. (b) ROI around the
645 mouse’s eye. (c) Histogram of pixel intensities in the ROI showing separation between the intensities in
646 the pupil and sclera. A threshold (blue) is set for initial diameter estimation based on a maximum
647 likelihood estimate of the sclera intensity (MLE, orange). (d) Image of the pupil after has been
648 thresholded in Radon space and transformed back into image space. (e) Detected pupil superimposed on
649 video frame. (f) Example showing changes in pupil diameter, whisker motion, EMG power, ∆[HbT], and
650 CA1 LFP during awake, NREM, and REM periods.
21
651
652
653 Fig. 2 | Quantification of pupil diameter across arousal states
654 (a) Mean pupil diameter (mm) for each arousal state. Mean of each animal is a circle. Grey diamond and
655 error bars are mean ± std. across animals for each arousal state. (b) Mean pupil diameter (z-units) for each
656 arousal state. (c) Whisking-evoked and stimulus evoked increases in pupil diameter based on the 2
657 seconds prior to event onset. (d) Pre-whitened power spectrum of pupil diameter (z-units) during different
658 arousal states. Shading (c, d) denotes standard error of the mean. Error bars (a, b) denote standard
659 deviation. Statistical comparisons shown are between Rest and other states using a GLME model with
660 Bonferroni correction for multiple comparisons (10) * α < 0.005, **α < 0.001, ***α < 0.0001. See Table
661 1 and Table 2 for additional statistical comparisons between each arousal state in (a, b).
662
22
663
664 Fig. 3 | Pupil diameter shows an arousal state dependent negative correlation with blood volume
665 and gamma-band power
666 (a-f) Relationship between gamma-band power and pupil diameter. (a) 2-D histogram showing the
667 relationship between pupil diameter and gamma-band power during the three arousal state classes. (b)
668 Cross-correlation between gamma-band power and pupil diameter for short duration arousal states. (c)
669 Cross-correlation for longer duration arousal states. (d) Coherence between gamma-band power and pupil
670 diameter. (e) Coherence at 0.02 Hz between gamma-band power and pupil diameter. (f) Coherence at
671 0.35 Hz. (g-l) Relationship between ∆[HbT] and pupil diameter (g) 2-D histogram showing the
672 relationship between pupil diameter and blood volume during the three arousal state classes. (h) Cross-
673 correlation between pupil diameter and blood volume for short duration arousal states. (i) Cross-
674 correlation for long duration arousal states. (j) Coherence between pupil diameter and blood volume. (k)
675 Coherence at 0.02 Hz. (l) Coherence at 0.35 Hz. Shading (b, c, h, i) denotes standard error of the mean.
676 Error bars (d, e, f, j, k, l) denote standard deviation. Statistics comparisons shown are between Rest/Alert
677 and other arousal states using a GLME mode with Bonferroni correction for multiple comparisons (3 in e,
678 k *α < 0.017, **α < 0.003, ***α < 0.0003 or 15 in f, l, *α < 0.003, **α < 0.00067, ***α < 0.000067, n.s.
679 – not significant). See Table 3, Table 4, Table 5, and Table 6 for additional statistical comparisons
680 between each arousal state in (e, f, k, l)
23
681
682 Fig. 4 | Blink-triggered neural and blood volume responses show arousal state-dependent changes
683 (a) Histogram of inter-blink interval for all blinks. (b) Mean inter-blink interval from each animal. (c)
684 Histogram of post-stim probability of a blink for whisker stimulations that elicited a blink. (Inset) Mean
685 probability from each animal of blinking after a whisker stimulus. (d) Peri-blink arousal states. (e) Peri-
686 blink probability of whisking during ‘Awake’ or ‘Asleep’ periods. (f-l) Blink-triggered averages for
687 blinks that occurred during an ‘Awake’ period, separated into either high whisk blinks (HWB) or low
688 whisk blinks (LWB). Periods where the pupil is partially obscured by the eyelid have been censored
24
689 (pink) (f) Pupil diameter during ‘Awake’ blinks (g) Hemodynamics during ‘Awake’ blinks (h)
690 Electromyography during ‘Awake’ blinks. (i) Cortical LFP during ‘Awake’ low-whisk blinks (LWB). (j)
691 Cortical LFP during ‘Awake’ high whisk blinks (HWB). (k) Hippocampal LFP during ‘Awake’ LWBs.
692 (l) Hippocampal LFP during ‘Awake’ HWBs. (m-s) Blink-triggered averages for blinks that occurred
693 during an ‘Asleep’ period, separated into either HWB or LWB events. (m) Pupil diameter during ‘Asleep’
694 blinks (n) Hemodynamics during ‘Asleep’ blinks (o) Electromyography during ‘Asleep’ blinks. (p)
695 Cortical LFP during ‘Asleep’ blinks with low amounts of whisking. (q) Cortical LFP during ‘Asleep’
696 blinks with high amounts of whisking. (r) Hippocampal LFP during ‘Asleep’ LWBs. (s) Hippocampal
697 LFP during ‘Asleep’ HWBs. Scatter plot error bars (b, c inset) are standard deviation. Shading (c main,
698 e, f, g, h, m, n, o) denotes standard error of the mean.
25
699
700
701 Fig. 5 | Effects of blinking on neural and bilateral hemodynamic signals
702 (a) Coherence between left and right gamma-band power before and after an awake blink. (b) Mean
703 coherence from 0 to 0.2 Hz before and after an awake blink. (c) Power spectrum of gamma-band power
704 envelope before and after an awake blink. (d) Mean power between 0 to 0.2 Hz before and after an awake
705 blink. (e) Change in pre- vs. post-blink gamma-band power vs. change in pre- vs. post-blink gamma-band
706 coherence for awake blinks. (f) Coherence between left and right ∆[HbT] before and after and awake
707 blink. (g) Mean coherence between left and right ∆[HbT] signals between 0 to 0.2 Hz before and after an
708 awake blink. (h) ∆[HbT] power before and after an awake blink. (i) Mean Power between 0 to 0.2 Hz
709 before and after an awake blink. (j) Change in pre- vs. post-blink power vs. change in pre- vs. post-blink
710 coherence for awake blinks. (k) Coherence between bilateral gamma-band signals before and after an
711 asleep blink. (l) Mean coherence between bilateral gamma-band signals between 0 to 0.2 Hz before and
712 after an asleep blink. (m) Power spectrum of the gamma-band power envelope before and after an asleep
713 blink (n) Mean power in the gamma band envelope between 0 to 0.2 Hz before and after an asleep blink.
714 (o) Change in pre- vs. post-blink power vs. change in pre- vs. post-blink coherence for asleep blinks. (p-t)
715 ∆[HbT] before and after an asleep blink (p) Coherence between bilateral hemodynamic signals before and
716 after a blink during sleep. (q) Mean coherence between 0 to 0.2 Hz before and after an asleep blink. (r)
26
717 ∆[HbT] power before and after an asleep blink. (s) Mean power between 0 to 0.2 Hz before and after an
718 asleep blink. (t) Change in pre- vs. post-blink power vs. change in pre- vs. post-blink coherence for asleep
719 blinks. Error bars (b, d, g, i, l, n, q, s) are standard deviation. Shading in (a, c, f, h, k, m, p, r) denote
720 standard error of the mean. Paired t-test *α < 0.05, **α < 0.01, ***α < 0.001, n.s. – not significant.
721
27
722
723
724 Fig. 6 | Pupil size, position, and motion change with arousal state
725 (a) Probability of arousal state classification as a function of pupil diameter (b) Diagram demonstrating
726 the drift of the pupil as the animal falls to sleep. The position of the centroid is measured relative to the
727 resting location. (c) Mean absolute velocity of the eye in each arousal state. (d-g) Pupil diameter (top) and
728 pupil position (bottom) as the animal transitions from one arousal state to another. During REM sleep, the
729 eye rotated towards the nose (nasally) and ventrally. Error bars and shading denote standard deviations.
730 Bonferroni corrected (3) paired t-test *α < 0.017, **α < 0.003, ***α < 0.0003, n.s. – not significant.
28
731
732 Fig. 7 | Eye metrics alone are an accurate predictor of arousal state
733 (a) Arousal state predictions from 3 arousal state classification models (Eye, Physiological, Combined)
734 and manual scores for sample data. Data is the same as presented in Figure 1. (b) Example trace of pupil
735 diameter over time with respect to arousal state. For clarity, blinks are not shown. (c) Position of the pupil
736 changes when the animal goes into ‘REM’ sleep, moving nasally and ventrally. (d) Absolute change in
737 the pupil’s frame-by-frame position indicates a high degree of movement during ‘REM’ sleep. (e)
738 Confusion matrix of a bagged random forest (70:30% training:testing) using only eye metrics such as
739 pupil diameter, position, and motion (‘Eye Model’) (f) Confusion matrix of a bagged random forest
740 (70:30% training:testing) using measurements of cortical and hippocampal LFP, electromyography, heart
741 rate, and whisker motion (‘Physiological Model’). (g) Confusion matrix of a model using both the eye
742 and physiological data (‘Combined Model’) for arousal scoring. (h) Out-of-bag error estimation during
743 model training for each model. Error bars in (h) denote standard deviation. Bonferroni corrected (3)
744 paired t-test *α < 0.017, **α < 0.003, ***α < 0.0003, n.s. – not significant.
29
745
746
747 Fig. 8 | Pupil diameter is anti-correlated with cortical hemodynamics during Rest and NREM sleep
748 Summary schematic demonstrating the observed correlation between pupil diameter and cortical
749 hemodynamics during each arousal state. Correlations between pupil diameter and blood volume are
750 positive during REM sleep and when the mouse is alert, moving, or stimulated. Correlations between
751 pupil diameter and blood volume are negative during rest and NREM sleep. While the oscillation
752 amplitude and baseline offset of the pupil diameter/blood volume reflect what is observed in our data, for
753 clarity the temporal frequencies of oscillations are not to scale.
754
755 Video 1. Video showing pupil diameter variations across several arousal states. Detected pupil area is in
756 purple.
30
References
Adams, M.D., Winder, A.T., Blinder, P., and Drew, P.J. (2018). The pial vasculature of the mouse
develops according to a sensory-independent program. Sci Rep 8, 9860. 10.1038/s41598-018-27910-3.
Amzica, F., and Steriade, M. (1998). Electrophysiological correlates of sleep delta waves.
Electroencephalogr Clin Neurophysiol 107, 69-83. 10.1016/s0013-4694(98)00051-0.
Bekar, L.K., Wei, H.S., and Nedergaard, M. (2012). The locus coeruleus-norepinephrine network
optimizes coupling of cerebral blood volume with oxygen demand. J Cereb Blood Flow Metab 32, 2135-
2145. 10.1038/jcbfm.2012.115.
Bellesi, M., Tononi, G., Cirelli, C., and Serra, P.A. (2016). Region-Specific Dissociation between
Cortical Noradrenaline Levels and the Sleep/Wake Cycle. Sleep 39, 143-154. 10.5665/sleep.5336.
Bergel, A., Deffieux, T., Demene, C., Tanter, M., and Cohen, I. (2018). Local hippocampal fast gamma
rhythms precede brain-wide hyperemic patterns during spontaneous rodent REM sleep. Nat Commun 9,
5364. 10.1038/s41467-018-07752-3.
Berridge, C.W., and Abercrombie, E.D. (1999). Relationship between locus coeruleus discharge rates and
rates of norepinephrine release within neocortex as assessed by in vivo microdialysis. Neuroscience 93,
1263-1270. Doi 10.1016/S0306-4522(99)00276-6.
Bokil, H., Andrews, P., Kulkarni, J.E., Mehta, S., and Mitra, P.P. (2010). Chronux: a platform for
analyzing neural signals. J Neurosci Methods 192, 146-151. 10.1016/j.jneumeth.2010.06.020.
Bradley, M.M., Miccoli, L., Escrig, M.A., and Lang, P.J. (2008). The pupil as a measure of emotional
arousal and autonomic activation. Psychophysiology 45, 602-607. 10.1111/j.1469-8986.2008.00654.x.
Breuninger, T., Puller, C., Haverkamp, S., and Euler, T. (2011). Chromatic bipolar cell pathways in the
mouse retina. J Neurosci 31, 6504-6517. 10.1523/JNEUROSCI.0616-11.2011.
Bristow, D., Frith, C., and Rees, G. (2005a). Two distinct neural effects of blinking on human visual
processing. Neuroimage 27, 136-145. 10.1016/j.neuroimage.2005.03.037.
Bristow, D., Haynes, J.D., Sylvester, R., Frith, C.D., and Rees, G. (2005b). Blinking suppresses the neural
response to unchanging retinal stimulation. Curr Biol 15, 1296-1300. 10.1016/j.cub.2005.06.025.
Burlingham, C.S., Mirbagheri, S., and Heeger, D.J. (2022). A unified model of the task-evoked pupil
response. Sci Adv 8, eabi9979. 10.1126/sciadv.abi9979.
Cantero, J.L., Atienza, M., Madsen, J.R., and Stickgold, R. (2004). Gamma EEG dynamics in neocortex
and hippocampus during human wakefulness and sleep. Neuroimage 22, 1271-1280.
10.1016/j.neuroimage.2004.03.014.
Cardoso, M.M.B., Lima, B., Sirotin, Y.B., and Das, A. (2019). Task-related hemodynamic responses are
modulated by reward and task engagement. PLoS Biol 17, e3000080. 10.1371/journal.pbio.3000080.
Carter, M.E., Yizhar, O., Chikahisa, S., Nguyen, H., Adamantidis, A., Nishino, S., Deisseroth, K., and de
Lecea, L. (2010). Tuning arousal with optogenetic modulation of locus coeruleus neurons. Nat Neurosci
13, 1526-1533. 10.1038/nn.2682.
31
Chang, L., Breuninger, T., and Euler, T. (2013). Chromatic coding from cone-type unselective circuits in
the mouse retina. Neuron 77, 559-571. 10.1016/j.neuron.2012.12.012.
Cirelli, C. (2009). The genetic and molecular regulation of sleep: from fruit flies to humans. Nat Rev
Neurosci 10, 549-560. 10.1038/nrn2683.
Drew, P., Sayres, R., Watanabe, K., and Shimojo, S. (2001). Pupillary response to chromatic flicker. Exp
Brain Res 136, 256-262. 10.1007/s002210000605.
Drew, P.J., and Feldman, D.E. (2009). Intrinsic signal imaging of deprivation-induced contraction of
whisker representations in rat somatosensory cortex. Cereb Cortex 19, 331-348. 10.1093/cercor/bhn085.
Drew, P.J., Mateo, C., Turner, K.L., Yu, X., and Kleinfeld, D. (2020). Ultra-slow Oscillations in fMRI
and Resting-State Connectivity: Neuronal and Vascular Contributions and Technical Confounds. Neuron
107, 782-804. 10.1016/j.neuron.2020.07.020.
Drew, P.J., Shih, A.Y., Driscoll, J.D., Knutsen, P.M., Blinder, P., Davalos, D., Akassoglou, K., Tsai, P.S.,
and Kleinfeld, D. (2010). Chronic optical access through a polished and reinforced thinned skull. Nat
Methods 7, 981-984. 10.1038/nmeth.1530.
Drew, P.J., Winder, A.T., and Zhang, Q. (2019). Twitches, Blinks, and Fidgets: Important Generators of
Ongoing Neural Activity. Neuroscientist 25, 298-313. 10.1177/1073858418805427.
Echagarruga, C.T., Gheres, K.W., Norwood, J.N., and Drew, P.J. (2020). nNOS-expressing interneurons
control basal and behaviorally evoked arterial dilation in somatosensory cortex of mice. Elife 9.
10.7554/eLife.60533.
Einhauser, W., Koch, C., and Carter, O.L. (2010). Pupil dilation betrays the timing of decisions. Front
Hum Neurosci 4, 18. 10.3389/fnhum.2010.00018.
Feng, J., Zhang, C., Lischinsky, J.E., Jing, M., Zhou, J., Wang, H., Zhang, Y., Dong, A., Wu, Z., Wu, H.,
et al. (2019). A Genetically Encoded Fluorescent Sensor for Rapid and Specific In Vivo Detection of
Norepinephrine. Neuron 102, 745-761 e748. 10.1016/j.neuron.2019.02.037.
Fultz, N.E., Bonmassar, G., Setsompop, K., Stickgold, R.A., Rosen, B.R., Polimeni, J.R., and Lewis, L.D.
(2019). Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep.
Science 366, 628-631. 10.1126/science.aax5440.
Gao, Y.R., and Drew, P.J. (2014). Determination of vessel cross-sectional area by thresholding in Radon
space. J Cereb Blood Flow Metab 34, 1180-1187. 10.1038/jcbfm.2014.67.
Gawne, T.J., and Martin, J.M. (2000). Activity of primate V1 cortical neurons during blinks. J
Neurophysiol 84, 2691-2694. 10.1152/jn.2000.84.5.2691.
Gilzenrat, M.S., Nieuwenhuis, S., Jepma, M., and Cohen, J.D. (2010). Pupil diameter tracks changes in
control state predicted by the adaptive gain theory of locus coeruleus function. Cogn Affect Behav
Neurosci 10, 252-269. 10.3758/CABN.10.2.252.
Goadsby, P.J., Lambert, G.A., and Lance, J.W. (1985). The mechanism of cerebrovascular
vasoconstriction in response to locus coeruleus stimulation. Brain Res 326, 213-217. 10.1016/0006-
8993(85)90030-7.
32
Golan, T., Davidesco, I., Meshulam, M., Groppe, D.M., Megevand, P., Yeagle, E.M., Goldfinger, M.S.,
Harel, M., Melloni, L., Schroeder, C.E., et al. (2016). Human intracranial recordings link suppressed
transients rather than 'filling-in' to perceptual continuity across blinks. Elife 5. 10.7554/eLife.17243.
Gray, S.R., Ye, L., Ye, J.Y., and Paukert, M. (2021). Noradrenergic terminal short-term potentiation
enables modality-selective integration of sensory input and vigilance state. Sci Adv 7, eabk1378.
10.1126/sciadv.abk1378.
Guipponi, O., Odouard, S., Pinede, S., Wardak, C., and Ben Hamed, S. (2015). fMRI Cortical Correlates
of Spontaneous Eye Blinks in the Nonhuman Primate. Cereb Cortex 25, 2333-2345.
10.1093/cercor/bhu038.
Hakerem, G., and Sutton, S. (1966). Pupillary response at visual threshold. Nature 212, 485-486.
10.1038/212485a0.
Hamel, E. (2006). Perivascular nerves and the regulation of cerebrovascular tone. J Appl Physiol 100,
1059-1064. 10.1152/japplphysiol.00954.2005.
Hayat, H., Regev, N., Matosevich, N., Sales, A., Paredes-Rodriguez, E., Krom, A.J., Bergman, L., Li, Y.,
Lavigne, M., Kremer, E.J., et al. (2020). Locus coeruleus norepinephrine activity mediates sensory-
evoked awakenings from sleep. Sci Adv 6, eaaz4232. 10.1126/sciadv.aaz4232.
Hess, E.H., and Polt, J.M. (1964). Pupil Size in Relation to Mental Activity during Simple Problem-
Solving. Science 143, 1190-1192. 10.1126/science.143.3611.1190.
Holland, M.K., and Tarlow, G. (1972). Blinking and mental load. Psychol Rep 31, 119-127.
10.2466/pr0.1972.31.1.119.
Huo, B.X., Gao, Y.R., and Drew, P.J. (2015). Quantitative separation of arterial and venous cerebral
blood volume increases during voluntary locomotion. Neuroimage 105, 369-379.
10.1016/j.neuroimage.2014.10.030.
Huo, B.X., Smith, J.B., and Drew, P.J. (2014). Neurovascular coupling and decoupling in the cortex
during voluntary locomotion. J Neurosci 34, 10975-10981. 10.1523/JNEUROSCI.1369-14.2014.
Hupe, J.M., Bordier, C., and Dojat, M. (2012). A BOLD signature of eyeblinks in the visual cortex.
Neuroimage 61, 149-161. 10.1016/j.neuroimage.2012.03.001.
International Brain, L., Aguillon-Rodriguez, V., Angelaki, D., Bayer, H., Bonacchi, N., Carandini, M.,
Cazettes, F., Chapuis, G., Churchland, A.K., Dan, Y., et al. (2021). Standardized and reproducible
measurement of decision-making in mice. Elife 10. 10.7554/eLife.63711.
Jing, M., Li, Y., Zeng, J., Huang, P., Skirzewski, M., Kljakic, O., Peng, W., Qian, T., Tan, K., Zou, J., et
al. (2020). An optimized acetylcholine sensor for monitoring in vivo cholinergic activity. Nat Methods
17, 1139-1146. 10.1038/s41592-020-0953-2.
Jones, B.E. (2020). Arousal and sleep circuits. Neuropsychopharmacology 45, 6-20. 10.1038/s41386-019-
0444-2.
33
Joshi, S., Li, Y., Kalwani, R.M., and Gold, J.I. (2016). Relationships between Pupil Diameter and
Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex. Neuron 89, 221-234.
10.1016/j.neuron.2015.11.028.
Kahneman, D., and Beatty, J. (1966). Pupil diameter and load on memory. Science 154, 1583-1585.
10.1126/science.154.3756.1583.
Kaminer, J., Powers, A.S., Horn, K.G., Hui, C., and Evinger, C. (2011). Characterizing the spontaneous
blink generator: an animal model. J Neurosci 31, 11256-11267. 10.1523/JNEUROSCI.6218-10.2011.
Karimi Abadchi, J., Nazari-Ahangarkolaee, M., Gattas, S., Bermudez-Contreras, E., Luczak, A.,
McNaughton, B.L., and Mohajerani, M.H. (2020). Spatiotemporal patterns of neocortical activity around
hippocampal sharp-wave ripples. Elife 9. 10.7554/eLife.51972.
Karson, C.N. (1979). Oculomotor signs in a psychiatric population: a preliminary report. Am J Psychiatry
136, 1057-1060. 10.1176/ajp.136.8.1057.
Karson, C.N. (1983). Spontaneous eye-blink rates and dopaminergic systems. Brain 106 (Pt 3), 643-653.
10.1093/brain/106.3.643.
Karson, C.N., Dykman, R.A., and Paige, S.R. (1990). Blink rates in schizophrenia. Schizophr Bull 16,
345-354. 10.1093/schbul/16.2.345.
Kedarasetti, R.T., Drew, P.J., and Costanzo, F. (2020a). Arterial pulsations drive oscillatory flow of CSF
but not directional pumping. Sci Rep 10, 10102. 10.1038/s41598-020-66887-w.
Kedarasetti, R.T., Drew, P.J., and Costanzo, F. (2022). Arterial vasodilation drives convective fluid flow
in the brain: a poroelastic model. Fluids Barriers CNS 19, 34. 10.1186/s12987-022-00326-y.
Kedarasetti, R.T., Turner, K.L., Echagarruga, C., Gluckman, B.J., Drew, P.J., and Costanzo, F. (2020b).
Functional hyperemia drives fluid exchange in the paravascular space. Fluids Barriers CNS 17, 52.
10.1186/s12987-020-00214-3.
Kim, J.H., Jung, A.H., Jeong, D., Choi, I., Kim, K., Shin, S., Kim, S.J., and Lee, S.H. (2016). Selectivity
of Neuromodulatory Projections from the Basal Forebrain and Locus Ceruleus to Primary Sensory
Cortices. J Neurosci 36, 5314-5327. 10.1523/JNEUROSCI.4333-15.2016.
Kjaerby, C., Andersen, M., Hauglund, N., Untiet, V., Dall, C., Sigurdsson, B., Ding, F., Feng, J., Li, Y.,
Weikop, P., et al. (2022). Memory-enhancing properties of sleep depend on the oscillatory amplitude of
norepinephrine. Nat Neurosci 25, 1059-1070. 10.1038/s41593-022-01102-9.
Kleinfeld, D., Blinder, P., Drew, P.J., Driscoll, J.D., Muller, A., Tsai, P.S., and Shih, A.Y. (2011). A
guide to delineate the logic of neurovascular signaling in the brain. Front Neuroenergetics 3, 1.
10.3389/fnene.2011.00001.
Larsen, R.S., and Waters, J. (2018). Neuromodulatory Correlates of Pupil Dilation. Front Neural Circuits
12, 21. 10.3389/fncir.2018.00021.
Le Van Quyen, M., Staba, R., Bragin, A., Dickson, C., Valderrama, M., Fried, I., and Engel, J. (2010).
Large-scale microelectrode recordings of high-frequency gamma oscillations in human cortex during
sleep. J Neurosci 30, 7770-7782. 10.1523/JNEUROSCI.5049-09.2010.
34
Liu, X., de Zwart, J.A., Scholvinck, M.L., Chang, C., Ye, F.Q., Leopold, D.A., and Duyn, J.H. (2018).
Subcortical evidence for a contribution of arousal to fMRI studies of brain activity. Nat Commun 9, 395.
10.1038/s41467-017-02815-3.
Liu, X., Yanagawa, T., Leopold, D.A., Chang, C., Ishida, H., Fujii, N., and Duyn, J.H. (2015). Arousal
transitions in sleep, wakefulness, and anesthesia are characterized by an orderly sequence of cortical
events. Neuroimage 116, 222-231. 10.1016/j.neuroimage.2015.04.003.
Longden, T.A., Dabertrand, F., Koide, M., Gonzales, A.L., Tykocki, N.R., Brayden, J.E., Hill-Eubanks,
D., and Nelson, M.T. (2017). Capillary K(+)-sensing initiates retrograde hyperpolarization to increase
local cerebral blood flow. Nat Neurosci 20, 717-726. 10.1038/nn.4533.
Ma, Y., Shaik, M.A., Kim, S.H., Kozberg, M.G., Thibodeaux, D.N., Zhao, H.T., Yu, H., and Hillman,
E.M. (2016a). Wide-field optical mapping of neural activity and brain haemodynamics: considerations
and novel approaches. Philos Trans R Soc Lond B Biol Sci 371. 10.1098/rstb.2015.0360.
Ma, Y., Shaik, M.A., Kozberg, M.G., Kim, S.H., Portes, J.P., Timerman, D., and Hillman, E.M. (2016b).
Resting-state hemodynamics are spatiotemporally coupled to synchronized and symmetric neural activity
in excitatory neurons. Proc Natl Acad Sci U S A 113, E8463-E8471. 10.1073/pnas.1525369113.
McGinley, M.J., David, S.V., and McCormick, D.A. (2015). Cortical Membrane Potential Signature of
Optimal States for Sensory Signal Detection. Neuron 87, 179-192. 10.1016/j.neuron.2015.05.038.
Megemont, M., McBurney-Lin, J., and Yang, H. (2022). Pupil diameter is not an accurate real-time
readout of locus coeruleus activity. Elife 11. 10.7554/eLife.70510.
Mirg, S., Chen, H., Turner, K.L., Gheres, K.W., Liu, J., Gluckman, B.J., Drew, P.J., and Kothapalli, S.R.
(2022a). Awake mouse brain photoacoustic and optical imaging through a transparent ultrasound cranial
window. Opt Lett 47, 1121-1124. 10.1364/OL.450648.
Mirg, S., Turner, K.L., Chen, H., Drew, P.J., and Kothapalli, S.R. (2022b). Photoacoustic imaging for
microcirculation. Microcirculation, e12776. 10.1111/micc.12776.
Montgomery, S.M., Sirota, A., and Buzsaki, G. (2008). Theta and gamma coordination of hippocampal
networks during waking and rapid eye movement sleep. J Neurosci 28, 6731-6741.
10.1523/JNEUROSCI.1227-08.2008.
Morad, Y., Lemberg, H., Yofe, N., and Dagan, Y. (2000). Pupillography as an objective indicator of
fatigue. Curr Eye Res 21, 535-542.
Murphy, P.R., O'Connell, R.G., O'Sullivan, M., Robertson, I.H., and Balsters, J.H. (2014). Pupil diameter
covaries with BOLD activity in human locus coeruleus. Hum Brain Mapp 35, 4140-4154.
10.1002/hbm.22466.
Musall, S., Kaufman, M.T., Juavinett, A.L., Gluf, S., and Churchland, A.K. (2019). Single-trial neural
dynamics are dominated by richly varied movements. Nat Neurosci 22, 1677-1686. 10.1038/s41593-019-
0502-4.
Nakamori, K., Odawara, M., Nakajima, T., Mizutani, T., and Tsubota, K. (1997). Blinking is controlled
primarily by ocular surface conditions. American Journal of Ophthalmology 124, 24-30. Doi
10.1016/S0002-9394(14)71639-3.
35
Nakano, T., Kato, M., Morito, Y., Itoi, S., and Kitazawa, S. (2013). Blink-related momentary activation
of the default mode network while viewing videos. Proc Natl Acad Sci U S A 110, 702-706.
10.1073/pnas.1214804110.
Nassar, M.R., Rumsey, K.M., Wilson, R.C., Parikh, K., Heasly, B., and Gold, J.I. (2012). Rational
regulation of learning dynamics by pupil-linked arousal systems. Nat Neurosci 15, 1040-1046.
10.1038/nn.3130.
Onorati, F., Barbieri, R., Mauri, M., Russo, V., and Mainardi, L. (2013). Characterization of affective
states by pupillary dynamics and autonomic correlates. Front Neuroeng 6, 9. 10.3389/fneng.2013.00009.
Osorio-Forero, A., Cardis, R., Vantomme, G., Guillaume-Gentil, A., Katsioudi, G., Devenoges, C.,
Fernandez, L.M.J., and Luthi, A. (2021). Noradrenergic circuit control of non-REM sleep substates. Curr
Biol 31, 5009-5023 e5007. 10.1016/j.cub.2021.09.041.
Pais-Roldan, P., Takahashi, K., Sobczak, F., Chen, Y., Zhao, X., Zeng, H., Jiang, Y., and Yu, X. (2020).
Indexing brain state-dependent pupil dynamics with simultaneous fMRI and optical fiber calcium
recording. Proc Natl Acad Sci U S A 117, 6875-6882. 10.1073/pnas.1909937117.
Paukert, M., Agarwal, A., Cha, J., Doze, V.A., Kang, J.U., and Bergles, D.E. (2014). Norepinephrine
controls astroglial responsiveness to local circuit activity. Neuron 82, 1263-1270.
10.1016/j.neuron.2014.04.038.
Petersen, C.C. (2007). The functional organization of the barrel cortex. Neuron 56, 339-355.
10.1016/j.neuron.2007.09.017.
Petersen, C.C. (2014). Cortical control of whisker movement. Annu Rev Neurosci 37, 183-203.
10.1146/annurev-neuro-062012-170344.
Pilorz, V., Tam, S.K., Hughes, S., Pothecary, C.A., Jagannath, A., Hankins, M.W., Bannerman, D.M.,
Lightman, S.L., Vyazovskiy, V.V., Nolan, P.M., et al. (2016). Melanopsin Regulates Both Sleep-
Promoting and Arousal-Promoting Responses to Light. PLoS Biol 14, e1002482.
10.1371/journal.pbio.1002482.
Pisauro, M.A., Benucci, A., and Carandini, M. (2016). Local and global contributions to hemodynamic
activity in mouse cortex. J Neurophysiol 115, 2931-2936. 10.1152/jn.00125.2016.
Pisauro, M.A., Dhruv, N.T., Carandini, M., and Benucci, A. (2013). Fast hemodynamic responses in the
visual cortex of the awake mouse. J Neurosci 33, 18343-18351. 10.1523/JNEUROSCI.2130-13.2013.
Poe, G.R., Foote, S., Eschenko, O., Johansen, J.P., Bouret, S., Aston-Jones, G., Harley, C.W., Manahan-
Vaughan, D., Weinshenker, D., Valentino, R., et al. (2020). Locus coeruleus: a new look at the blue spot.
Nat Rev Neurosci 21, 644-659. 10.1038/s41583-020-0360-9.
Polack, P.O., Friedman, J., and Golshani, P. (2013). Cellular mechanisms of brain state-dependent gain
modulation in visual cortex. Nat Neurosci 16, 1331-1339. 10.1038/nn.3464.
Preuschoff, K., t Hart, B.M., and Einhauser, W. (2011). Pupil Dilation Signals Surprise: Evidence for
Noradrenaline's Role in Decision Making. Front Neurosci 5, 115. 10.3389/fnins.2011.00115.
36
Privitera, M., Ferrari, K.D., von Ziegler, L.M., Sturman, O., Duss, S.N., Floriou-Servou, A., Germain,
P.L., Vermeiren, Y., Wyss, M.T., De Deyn, P.P., et al. (2020). A complete pupillometry toolbox for real-
time monitoring of locus coeruleus activity in rodents. Nat Protoc 15, 2301-2320. 10.1038/s41596-020-
0324-6.
Raichle, M.E., Hartman, B.K., Eichling, J.O., and Sharpe, L.G. (1975). Central noradrenergic regulation
of cerebral blood flow and vascular permeability. Proc Natl Acad Sci U S A 72, 3726-3730.
10.1073/pnas.72.9.3726.
Rasmussen, R., Nicholas, E., Petersen, N.C., Dietz, A.G., Xu, Q., Sun, Q., and Nedergaard, M. (2019).
Cortex-wide Changes in Extracellular Potassium Ions Parallel Brain State Transitions in Awake Behaving
Mice. Cell Rep 28, 1182-1194 e1184. 10.1016/j.celrep.2019.06.082.
Reimer, J., Froudarakis, E., Cadwell, C.R., Yatsenko, D., Denfield, G.H., and Tolias, A.S. (2014). Pupil
fluctuations track fast switching of cortical states during quiet wakefulness. Neuron 84, 355-362.
10.1016/j.neuron.2014.09.033.
Reimer, J., McGinley, M.J., Liu, Y., Rodenkirch, C., Wang, Q., McCormick, D.A., and Tolias, A.S.
(2016). Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat
Commun 7, 13289. 10.1038/ncomms13289.
Salkoff, D.B., Zagha, E., McCarthy, E., and McCormick, D.A. (2020). Movement and Performance
Explain Widespread Cortical Activity in a Visual Detection Task. Cereb Cortex 30, 421-437.
10.1093/cercor/bhz206.
Sanchez-Lopez, A., and Escudero, M. (2011). Tonic and phasic components of eye movements during
REM sleep in the rat. Eur J Neurosci 33, 2129-2138. 10.1111/j.1460-9568.2011.07702.x.
Saper, C.B., and Fuller, P.M. (2017). Wake-sleep circuitry: an overview. Curr Opin Neurobiol 44, 186-
192. 10.1016/j.conb.2017.03.021.
Schwarz, L.A., and Luo, L. (2015). Organization of the locus coeruleus-norepinephrine system. Curr Biol
25, R1051-R1056. 10.1016/j.cub.2015.09.039.
Senzai, Y., and Scanziani, M. (2022). A cognitive process occurring during sleep is revealed by rapid eye
movements. Science 377, 999-1004. 10.1126/science.abp8852.
Shih, A.Y., Mateo, C., Drew, P.J., Tsai, P.S., and Kleinfeld, D. (2012). A polished and reinforced
thinned-skull window for long-term imaging of the mouse brain. J Vis Exp. 10.3791/3742.
Siegle, G.J., Ichikawa, N., and Steinhauer, S. (2008). Blink before and after you think: blinks occur prior
to and following cognitive load indexed by pupillary responses. Psychophysiology 45, 679-687.
10.1111/j.1469-8986.2008.00681.x.
Sirotin, Y.B., and Das, A. (2009). Anticipatory haemodynamic signals in sensory cortex not predicted by
local neuronal activity. Nature 457, 475-479. 10.1038/nature07664.
Sobczak, F., Pais-Roldan, P., Takahashi, K., and Yu, X. (2021). Decoding the brain state-dependent
relationship between pupil dynamics and resting state fMRI signal fluctuation. Elife 10.
10.7554/eLife.68980.
37
Steriade, M., Nunez, A., and Amzica, F. (1993). A novel slow (< 1 Hz) oscillation of neocortical neurons
in vivo: depolarizing and hyperpolarizing components. The Journal of Neuroscience 13, 3252-3265.
10.1523/jneurosci.13-08-03252.1993.
Stern, J.A., Boyer, D., and Schroeder, D. (1994). Blink rate: a possible measure of fatigue. Hum Factors
36, 285-297. 10.1177/001872089403600209.
Stern, J.A., Walrath, L.C., and Goldstein, R. (1984). The endogenous eyeblink. Psychophysiology 21, 22-
33. 10.1111/j.1469-8986.1984.tb02312.x.
Strauch, C., Wang, C.A., Einhauser, W., Van der Stigchel, S., and Naber, M. (2022). Pupillometry as an
integrated readout of distinct attentional networks. Trends Neurosci. 10.1016/j.tins.2022.05.003.
Stringer, C., Pachitariu, M., Steinmetz, N., Reddy, C.B., Carandini, M., and Harris, K.D. (2019).
Spontaneous behaviors drive multidimensional, brainwide activity. Science 364, 255.
10.1126/science.aav7893.
Sullivan, D., Mizuseki, K., Sorgi, A., and Buzsaki, G. (2014). Comparison of sleep spindles and theta
oscillations in the hippocampus. J Neurosci 34, 662-674. 10.1523/JNEUROSCI.0552-13.2014.
Tagliazucchi, E., and Laufs, H. (2014). Decoding wakefulness levels from typical fMRI resting-state data
reveals reliable drifts between wakefulness and sleep. Neuron 82, 695-708.
10.1016/j.neuron.2014.03.020.
Toussay, X., Basu, K., Lacoste, B., and Hamel, E. (2013). Locus coeruleus stimulation recruits a broad
cortical neuronal network and increases cortical perfusion. J Neurosci 33, 3390-3401.
10.1523/JNEUROSCI.3346-12.2013.
Tran, C.H.T., Peringod, G., and Gordon, G.R. (2018). Astrocytes Integrate Behavioral State and Vascular
Signals during Functional Hyperemia. Neuron 100, 1133-1148 e1133. 10.1016/j.neuron.2018.09.045.
Turner, K.L., Gheres, K.W., Proctor, E.A., and Drew, P.J. (2020). Neurovascular coupling and bilateral
connectivity during NREM and REM sleep. Elife 9. 10.7554/eLife.62071.
Van Orden, K.F., Limbert, W., Makeig, S., and Jung, T.P. (2001). Eye activity correlates of workload
during a visuospatial memory task. Hum Factors 43, 111-121. 10.1518/001872001775992570.
van Veluw, S.J., Hou, S.S., Calvo-Rodriguez, M., Arbel-Ornath, M., Snyder, A.C., Frosch, M.P.,
Greenberg, S.M., and Bacskai, B.J. (2020). Vasomotion as a Driving Force for Paravascular Clearance in
the Awake Mouse Brain. Neuron 105, 549-561 e545. 10.1016/j.neuron.2019.10.033.
Vazquez, A.L., Fukuda, M., Crowley, J.C., and Kim, S.G. (2014). Neural and hemodynamic responses
elicited by forelimb- and photo-stimulation in channelrhodopsin-2 mice: insights into the hemodynamic
point spread function. Cereb Cortex 24, 2908-2919. 10.1093/cercor/bht147.
Vinck, M., Batista-Brito, R., Knoblich, U., and Cardin, J.A. (2015). Arousal and locomotion make
distinct contributions to cortical activity patterns and visual encoding. Neuron 86, 740-754.
10.1016/j.neuron.2015.03.028.
Weber, F., and Dan, Y. (2016). Circuit-based interrogation of sleep control. Nature 538, 51-59.
10.1038/nature19773.
38
Winder, A.T., Echagarruga, C., Zhang, Q., and Drew, P.J. (2017). Weak correlations between
hemodynamic signals and ongoing neural activity during the resting state. Nat Neurosci 20, 1761-1769.
10.1038/s41593-017-0007-y.
Xie, L., Kang, H., Xu, Q., Chen, M.J., Liao, Y., Thiyagarajan, M., O'Donnell, J., Christensen, D.J.,
Nicholson, C., Iliff, J.J., et al. (2013). Sleep drives metabolite clearance from the adult brain. Science 342,
373-377. 10.1126/science.1241224.
Yang, H., Bari, B.A., Cohen, J.Y., and O'Connor, D.H. (2021). Locus coeruleus spiking differently
correlates with S1 cortex activity and pupil diameter in a tactile detection task. Elife 10.
10.7554/eLife.64327.
Yoss, R.E., Moyer, N.J., and Hollenhorst, R.W. (1970). Pupil size and spontaneous pupillary waves
associated with alertness, drowsiness, and sleep. Neurology 20, 545-554. 10.1212/wnl.20.6.545.
Yuzgec, O., Prsa, M., Zimmermann, R., and Huber, D. (2018). Pupil Size Coupling to Cortical States
Protects the Stability of Deep Sleep via Parasympathetic Modulation. Curr Biol 28, 392-400 e393.
10.1016/j.cub.2017.12.049.
Zhang, Q., Cramer, S.R., Ma, Z., Turner, K.L., Gheres, K.W., Liu, Y., Drew, P.J., and Zhang, N. (2022a).
Brain-wide ongoing activity is responsible for significant cross-trial BOLD variability. Cereb Cortex.
10.1093/cercor/bhac016.
Zhang, Q., Roche, M., Gheres, K.W., Chaigneau, E., Kedarasetti, R.T., Haselden, W.D., Charpak, S., and
Drew, P.J. (2019). Cerebral oxygenation during locomotion is modulated by respiration. Nat Commun 10,
5515. 10.1038/s41467-019-13523-5.
Zhang, Q., Turner, K.L., Gheres, K.W., Hossain, M.S., and Drew, P.J. (2022b). Behavioral and
physiological monitoring for awake neurovascular coupling experiments: a how-to guide. Neurophotonics
9, 021905. 10.1117/1.NPh.9.2.021905.
39

Relating Pupil Diameter and Blinking To Cortical Activity and Hemodynamics Across Arousal States

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Relating Pupil Diameter and Blinking To Cortical Activity and Hemodynamics Across Arousal States

Uploaded by

Copyright:

Available Formats

bioRxiv preprint doi: https://doi.org/10.1101/2022.06.15.496309; this version posted October 31, 2022.

The copyright holder for this preprint

170 bwconvhull, imfill, regionprops, diff, fillmissing.

398 Table 6 | Statistical comparisons of pupil-∆[HbT] coherence at 0.35 Hz

637 Fig. 1 | Pupil diameter tracks arousal state

You might also like